Originally Posted by will. There is a pattern amerging here! It is the pattern of everything... They releace something, its awesome, and i'm all set to spend money on it.. and now i have to wait for the next new thing to come out! Whats the bet that 2 months later they will announce another new thing and we'll be told to wait for that as well... I think sometimes we just have to take the plunge, upgrade and miss out a few of these advancements... Unless your uber rich like some of the people on this forum seem/have to be!
I'd normally agree with you here, but by chance, luck or a weird planet aligment, the first quarter (not fiscally speaking) of 2007 is going to be one of the best moments to upgrade ever. Waiting for 6 months or so will give you acces to Quad-core CPUs, DX10 GPUs and a new OS. When was the last time something similar happened?
Quote:
Originally Posted by aggies11 Conroe was a big jump, so it was worth waiting for. The benefits of Quad core will likely only be seen on paper.
Aggies
Likely, but what if the main benefit is a price drop in Conroe? Waiting a couple or 3 months for Conroe is a good advice anyway, if only so you get a good selection of motherboards (with perhaps more stable BIOS revisions) and new coolers to choose from. Even if I had to upgrade now, I'd wait until I see how nForce5 or RD600 based boards perform or what budget boards will be available. And IIRC, there's at least one new Conroe stepping coming.
So by the time the entire plataform is mature enough and availability and prices have stabilized, Kentsfield will be really near. I agree that the performance jump may not be huge compared with Conroe, but surely is much more sensible to wait a bit and have more options.
I know THG did a shoot out with a dual core EE CPU not too long ago (I wish I could remember the name of the CPU) so it had dual cores and hyper threading and was recognized as 4 CPUs in their XP machine. Maybe physical cores makes a difference?
Originally Posted by mattan Does Win XP Pro have support for 4 processors (cores)?
Microsoft defines 'processor' and 'core' differently. It is possible to run four threads in Windows XP - Intel's Pentium Extreme Edition 965, 955 and 840 chips had dual cores with hyperthreading technology (meaning a total of four threads).
Interesting... I may have to wait until next year now. In any case it might not be in my interest to upgrade just yet if my plans for this year go ahead.
4 cores should be noticeably fast by the time they are released. People now are beginning to write multi-threaded versions of their applications. In theory if they can write dual core compatible software it shouldn't be much harder to create 4 or 8 core compatible software. It's going from 1 thread to 2 threads which is the hard bit but from there, depending on the software, increasing the number of threads gets easier. We should see lots more massively parallel applications in the future.
Nice summary, but dont get everyone excited when it's something they will be very likley never get the chance to use.
Tbh I think that anyone waiting for Quad Core as a desktop platform would be making a big mistake.
Quad Core like Dual core is designed specifically for the server market. There are still only a few applications which are programmed to take advantage of Dual Core processors. The issue is that it cost a whole lot more to develop applications to run on Dual core processors, so just think about the cost for programming Quad core.
For normal users doing a little video encoding, gaming and surfing then Dual Core will be fine. Conroe will continue to evolve in its Dual Core incarnation and owning one will probably be more than enough.
Someone spoke the other day about the fact that were still on 32 bit technology despite so many people owning 64 bit processors. It's unlikley that we will see a complete migration of all applications to 64 bit within foreseable future. As for Quad Core as a viable desktop processor...forget about it. Fully 64 bit & Quad core are currently only viable in the Enterprise Server market, and is outside the realms of reality for 99% of the Computer Enthusiat market.
I agree with the point made about coders not writing for quad core until a good percentage of the market is using machines based on these processors, and the same can be said for the jump from 32 to 64 bit procs.
Personally, my next upgrade around xmas will be stepping up to a AM2 mobo and a AMD X2 chip... mainly because I've been running my machine in it's current config (3200+ Barton, 1GB DDR400, 9600XT) for going on 2 years and it has served me well even up to today (Go AMD!). I get nice frame rates and good visuals even with Elder Scrolls 4, though I will admit I run it at a lower resolution than I'd prefer in order to keep draw distances higher... the lack of HDR support on my gpu doesn't bother me, as I am planning an upgrade in the near future.
In essense, my point is that I have an outdated machine by most of your opinions, but it does everything a top-end rig will do almost as well. I see no practical purpose for an average user to step up to quad-core 64-bit chips, other than bragging rights for having a bleeding-edge rig.
That being said, as software devs make the change as the marketplace does, 6-12 months from now it will probably be a different story.
BTW, anybody else seeing Moore's Law packing it's bags?
Originally Posted by DeX n theory if they can write dual core compatible software it shouldn't be much harder to create 4 or 8 core compatible software.
It depends on the specific task. Many, if not most, applications arent' very parallelizable. It may take a great deal of effort/difficulty to split that task into only two evenly balanced workloads. Trying to go past that may be either next to impossible, or at least exponentially more difficult.
Certain tasks, for example image rendering, are extremely paralellizable, you can easily divy up the image into 8chunks, because what's in one part does not depend on the other. Few tasks are that accomodating though, as most have tons of linear dependencies and can't be split up that well.
It's not so much a matter of time, it's a matter of difficulty, and often impossibility.
Quad Core on the desktop gains will mostly be seen by the "System Idle Process" ;)
I think some people have misunderstood exactly what I've said here so I am going to attempt to clarify.
Quad-core is unlikely to make a big impact on the market when it first launches in exactly the same way that dual-core didn't, relatively speaking. However, since then, many apps have become multi-threaded - just about every test in our CPU benchmarks suite is multithreaded in one way or another.
The main point I made was slightly hidden, but basically I stand by my original opinion when I reviewed Core 2 Duo. There is little need to spend anymore than US$316 on a Core 2 Duo, because with that you're going to get flagship performance - at least, if you look at AMD's product stack after the price drops come into effect. Buying a flagship Core 2 Duo X6800 is just like buying an Athlon 64 FX-57 - there is no doubting that it is the fastest CPU out there at the moment, but it is soon to be replaced by something with more cores.
In the long run, those cores are going to get used - it's not taken long for developers to develop multithreaded code for dual core processors, so I don't think it will be much more than 12 months before we see massively-multithreaded applications on the market.
Originally Posted by aggies11 Certain tasks, for example image rendering, are extremely paralellizable, you can easily divy up the image into 8chunks, because what's in one part does not depend on the other. Few tasks are that accomodating though, as most have tons of linear dependencies and can't be split up that well.
Yes I guess you are right that some applications won't be able to fully utilise all the cores. I was thinking more of those applications that have been adapted to run with two cores - they are likely to run well with 4 or more cores.
But then when it comes to performance demanding applications such as games I'm sure we'll see that as developers gain access to more and more power they will find stuff to do to make use of it. For example bolting on more physics and AI where the power to do these tasks is available.
I've read several reports that the introduction of conroe isn't going to be sudden, more a gradual phasing in.
Also, there have been reports that a large portion of the first chips off the line will be going to the likes of DELL and APPLE.
Given the excitement over conroe and the fact that EVERYBODY wants to get one, and a LOT of people have been putting off upgrading till now, is it not possible that INTEL is slightly worried that with limited availability, and the massive price cuts by AMD, some people may be put off from waiting longer and plump for a nice cheap dual core athlon64 ready for the christmas holidays.
By getting everyone excited about something even better that's just around the corner, they can ensure more people will be persuaded to hold off on that upgrade, something they're desparate to do, until kentsfield arrives. By the time people realise that kentsfield will more than likely be out of the price range of many, production should have been able to catch up with the demand for conroe.
This is pure speculation, but there has to be a point to all of this, and INTEL still want to be able to cash in on the Christmas shoppers. If not enough conroes come out, surely, some buyers will be tempted by AMDs low prices.
not to mention, with the x360 already possessing triple cores, and the cell with its multiple SPE's (or whatever it is) programmers are getting used to adjusting their code writing habbits to make room for increased core count.
Ad to that intels implied strategy of no longer pushing the frequency envelope and instead pushing the thread count (8 core, and i read an article projecting that intel plans to have a 32 core processor available sometime circa 2010) we could very well see a progressive decrease in the turnaround time it takes software developers to include code which would take advantage of these chips.
The transition from single core to dual core was a unique development. Going from 2x core to 4x core is essentially the same thing.. so it only makes sense that the whole process takes less time. Gonig from 4x to 8x will see additional decrease in turn around time.. etc etc etc. I'm sure you see my point.
Also, another point. With physics and what not in games.. I'm sure games could put the extra cores to work. Granted, it wouldnt be as solid a solution as having a PPU such as Agieas offering.. as they handle data and tasks on completely different levels but 2x cores with physics computations being effectively thrown at it while another two deal with standard cpu functions and background tasks would hardly be something to scoff at.
I was planning on waiting until december and G80 to purchase a new cpu anyway.. as my 805D @ 3.7ghz does alright for me for the time being. This makes me feel good about my decision. What people SHOULD do is always a decision that is heavily effected by perspective and circumstance among many other factors, but waiting for kentsfield certainly is up there with one of the smarter of decisions for those who have upgrade cycles in the 1.5+ year range.
Originally Posted by 8-BALL OK, stretching it a bit here :)
I've read several reports that the introduction of conroe isn't going to be sudden, more a gradual phasing in.
Also, there have been reports that a large portion of the first chips off the line will be going to the likes of DELL and APPLE.
Given the excitement over conroe and the fact that EVERYBODY wants to get one, and a LOT of people have been putting off upgrading till now, is it not possible that INTEL is slightly worried that with limited availability, and the massive price cuts by AMD, some people may be put off from waiting longer and plump for a nice cheap dual core athlon64 ready for the christmas holidays.
By getting everyone excited about something even better that's just around the corner, they can ensure more people will be persuaded to hold off on that upgrade, something they're desparate to do, until kentsfield arrives. By the time people realise that kentsfield will more than likely be out of the price range of many, production should have been able to catch up with the demand for conroe.
This is pure speculation, but there has to be a point to all of this, and INTEL still want to be able to cash in on the Christmas shoppers. If not enough conroes come out, surely, some buyers will be tempted by AMDs low prices.
8-ball
Interesting thoughts 8-ball. Only time will tell with respect to Intel's availability, but quad-core is definitely something that Otellini is using to prevent users from heading off to AMD.
Originally Posted by specofdust I disagree. My main system is used primarily for gaming, but I'm shortly going to be building a PVR too. If I want, I'd be able to do the heavy duty encoding on a quad core core over the network, thus keeping the PVR low powered in every way, which has both silence and cost benefits. That's one core. Games are starting to be coded for two cores. True, it'll probably take time before 100% of two cores are used by games, but I have no doubt in the ability of programmers everywhere to abuse any CPU I have to the fullest. That's three cores. That leaves one core over for the usuals, scheduled virus scans, real time monitoring apps that lean heavily on the system, and all the other random things people do.
I agree that quad core wouldn't be for everyone, but as a main processor for all my PC's quad core sounds fantastic.
Lastly, if people don't build 4 core CPU's no-one will ever code for 4 core CPU's. These things have to be being used by people before the coders will bother, but you can be sure that they will and within a few years apps that need a lot of power will be able to utilise all that is available to them.
I hate to burst your bubble but if all of those applications are not built to use 4 cores....then most will be fighting for the first one while the other three will be idle.
I got linked to this review/news article. Its sad to see the disinformation that gets put out on the internet.
Some facts really quick
If you look at all of the applications that are currently on the market today for your PC, 95% of them are 32bit and support ONE CORE.
We have had 64bit hardware for over two years now and still the vast majority of applications are 32bit.
Even on platforms such as the 360 most games are NOT using more than a single core.
Waiting for a 4 core CPU to upgrade your personal computer is a total and utter joke.
I would wager than 1 year from now most people will still be running a 32bit operating system with a single core CPU and single threaded applications.
I will go so far as to say that a year from now anyone running Vista on a Conroe will still be running 70-80% of their applications that only support a single core and NONE of them will support more than 2 cores.
If you upgraded your rig to a Conroe and say a 7900GT with 2gigs of ram that it would last most people from 3-5 years. The PC gaming market is shrinking has been for the last 5 years. The only upgrade you might need in those 3 years would be a DX-10 video card if some killer game came out that did not support DX9 backward compatibility .because they only wanted a few sales.
There is so much bad advice on the Internet today.
Originally Posted by SGT Lindy I hate to burst your bubble but if all of those applications are not built to use 4 cores....then most will be fighting for the first one while the other three will be idle.
That is not quite true, I'm guessing you've never used a dual-core processor. Typically the threads are shared across the two cores if they're not multi-threaded. It's also possible to set affinity on specific threads in Windows Task Manager. The scheduling in Windows XP isn't the greatest, but Vista works much better in that respect, meaning that it's more friendly to multi core CPUs.
Quote:
Some facts really quick
If you look at all of the applications that are currently on the market today for your PC, 95% of them are 32bit and support ONE CORE.
We have had 64bit hardware for over two years now and still the vast majority of applications are 32bit.
Even on platforms such as the 360 most games are NOT using more than a single core.
Applications like Adobe Photoshop, Adobe Premiere, 3DS MAX, WinRAR, iTunes, FlaskMPEG, DVD Shrink, DivX Codec 6.1, MusicMatch Jukebox, TMPGEnc, VirtualDub, XMPEG, Windows Media Encoder, Xvid Codec 1.2, and the next version of Microsoft Office are all multi-threaded.
Games like Elder Scrolls IV: Oblivion, Quake 4, Call Of Duty 2, F.E.A.R., City Of Villians/Heroes and (I think) Prey are multi-threaded and more multi-threaded games are turning up.
These are just a few popular multi-threaded applications that I've listed off the top of my head.
Quote:
Waiting for a 4 core CPU to upgrade your personal computer is a total and utter joke.
Each to their own.
Quote:
I would wager than 1 year from now most people will still be running a 32bit operating system with a single core CPU and single threaded applications.
Depends whether you're talking about most of all computer users, or most enthusiasts. Considering that bit-tech is a publication who's average reader is a 24 year old male that spends a lot of money each year on computer hardware, we're targetted at the enthusiast sector.
Quote:
There is so much bad advice on the Internet today.
I'm gonna agree with everything you said, although a bit mildier ;)
The fact of the matter, switching to "multi-threaded development" is not simply just a paradigm shift in software development. It's not just "another way of doing things, that given more time you will get better with". There are some (many) tasks that will *never* be multithreaded. Not because programmers are lazy, or it's "new", because it's fundamentally impossible. Some tasks *can't* be broken up into multiple subtasks that can be executed at the same time. Some (many) things are just linear, and unless you can find a way to violate the laws of temporal physics (quantum computing anyone? ;) ), there is no way around it.
Multi-core/parallelization is not the Holy Grail of computing, it is not even "the next logical step". It exists for one reason only, COST. It is simply "cheaper to make 4 slower cores, than one single faster one". The cost and difficulty of developing faster single core chips has risen alot. It's just plain cheaper to stick multiple slower/"easier" cores in that same space. It isn't a perfect solution, because essentially hardware is "passing the buck" onto software.
And quite often, there is nothing software can do about it.
If anything, these facts are the reason why there was so much big fanfare about the rumoured "anti-hyperthreading" or "reverse-hyperthreading" from AMD. Because many things are gonna stay singlecore, and it'd be great to have multi-core "somehow" pretend to be singlecore. But that rumour was obviously wishfull thinking. It can't exist, because it's impossible. There are some problems there are just no ways around. To use a horrible/trivial example: if I want to do the math problem: (4 x 6) + 7. There is no way to break that into two steps that can be done at the same time. No matter how many cores you have, you have to wait for the result of "4 x 6", before you can add 7. These are fundamental laws of space time ;). Many aspects of computing are like that. You can't do the next step, until the results of the previous one are known. That means there is no way to parallelize it.
So yeah, don't get me wrong, if the choice is between 4cores and 2cores for roughly the same price, go for it. It can't hurt. But there is no sense waiting for something that won't see benefits for a long time, if at all. Conroe is real, and is much better. The gains of quad-core are largely theorhetical at this point, and could conceivably be for quite some time.
Originally Posted by SGT Lindy I hate to burst your bubble but if all of those applications are not built to use 4 cores....then most will be fighting for the first one while the other three will be idle.
You need to really have an idea of how an operating system uses multiple cores, this is entirely not the way. So, you start by claiming disinformation, yet, you dont understand it yourself.
95% of software is 32bit, but anyone can make up statistics, Kent, 64% of people know that.
Are we talking about software that was made in 1995 or software that doesnt need to be or wont benefit from being multithreaded? Having more cores enables people to do more at once, in the background, having a seemless, smoother computing experience. It's not the end answer by any means, but you it gives developers more to play with and people more to do as more of our lives become embedded in our computers. So with 4 cores you can now download, play a game, listen to music, transcode a DVD and burn another all at the same time. Should you want to get more things done in less time.
Quote:
We have had 64bit hardware for over two years now and still the vast majority of applications are 32bit.
USB and PCI took years to take hold. PCI Express has also taken a while. The next Vista will be avaliable in both and drivers are being made avalialbe for both. It's a chicken and egg situation but new CPUs support both, so if something benefits from being 64bit and you're running a 64bit OS, then you get things done faster.
Quote:
Even on platforms such as the 360 most games are NOT using more than a single core.
Sure, first produced games wont, but that's because they dont need it. That's the same with every single console and every new bit of hardware: the software has to have time to catch up, it wont always be the case. Future iterations will be produced and will need more processing power to do other stuff. There's room for expansion. You can't bring out an Xbox every year.
Quote:
I would wager than 1 year from now most people will still be running a 32bit operating system with a single core CPU and single threaded applications.
Im good for that bet? Whatcha want on it? ;) What are the definitions of most people? Most people on here or who read these types of reviews on the net? Most people who will be buying system from now until next year?
Quote:
I will go so far as to say that a year from now anyone running Vista on a Conroe will still be running 70-80% of their applications that only support a single core and NONE of them will support more than 2 cores.
Vista is being made multithreaded. Depends what they use and how many of those programs are running at once. Multithreading is distributed, it's not necessarily designed for a certain core quantity, depending on what it does.
Quote:
If you upgraded your rig to a Conroe and say a 7900GT with 2gigs of ram that it would last most people from 3-5 years.
2-3 years, 6-18 months for people on here. But you wont be buying a 7900GT if you aren't going to game on it. If you're just gonna surf the net then buy a mac mini. As larger TFTs come down in price you're gonna need to generate more pixels to fill that screen. I wonder how many people on here are running 3-5 year old graphics cards? 3 years is.. Geforce 4 range, 5 is.. Geforce 256? 2 GTS?
Quote:
The PC gaming market is shrinking has been for the last 5 years.
That's why it now annually grosses more than movies do and games like WoW have 6 million people playing? That the next generation of consoles is going to be the biggest ever and the graphics card evolution rate has quadrupled that of CPUs... because there's no money in it.
Quote:
There is so much bad advice on the Internet today.
That's because there are so many misinformed people ;) :):)
Applications like Adobe Photoshop, Adobe Premiere, 3DS MAX, WinRAR, iTunes, FlaskMPEG, DVD Shrink, DivX Codec 6.1, MusicMatch Jukebox, TMPGEnc, VirtualDub, XMPEG, Windows Media Encoder, Xvid Codec 1.2, and the next version of Microsoft Office are all multi-threaded.
To be fair, while I don't share the voracity of the previous poster( multi-core is definitely NOT a complete waste of time), I feel I have to point out a few things. That all the above apps fall into the category of "media encoding". And there are tons of overlap/redundancy on that list. Yes media encoding is the perfect candidate for parallelization, but it's only one small section of all the types of software that can/is run on a PC.
Quote:
Originally Posted by Tim S
Games like Elder Scrolls IV: Oblivion, Quake 4, Call Of Duty 2, F.E.A.R., City Of Villians/Heroes and (I think) Prey are multi-threaded and more multi-threaded games are turning up.
Multi-threaded is one thing, but I don't think we can say that the games have a balanced load between the cores. 2x2GHz does not perform in-game (CPU wise) relative to a 1x4GHz. There are benefits mind you, but not nearly in the range to say that the games are taking "full advantage" of both cores.
Again, if you offer me a nth-core chip right now, I'm surely gonna take it. But we shouldn't be under the misguided belief that multi-core gives the same benefits of faster single-core, "once the software devs have enough time to get used to it".
I wouldn't be surprised that if tomorrow, magically, 100% of software became as parallelized as possible, the gains would not be nearly as large as everyone expects (or rather, as we have all been lead to believe).
It never has done, just the same as buying twice as fast ram never equates to a twice as fast memory access. You will get diminishing returns but if we lived like that we'd still be running Pentium 90s.
Again, though, it's not necessarily about doing just one thing faster.
Aggies, it ultimately depends on how you use your computer. I used to run a single core processor and I would say I am a CPU-intensive computer user. I'm impatient and I want a smooth computing experience because I simply don't have the time to sit waiting for CPU-heavy tasks to complete.
My move from single core to dual core allowed me to do much more because I am no longer waiting for CPU-heavy tasks to complete. I can now burn a CD/DVD in the background, encode a DVD/video/audio stream in the background. I can work on high resolution images in photoshop without my PC slowing down. I find that the computing experience is completely different - it's hard to explain how and I am sure that many other dual core users will say the same.
Now, I am not expecting the overall computing experience to massively improve with quad-cores, but I am expecting a lot of applications to move to multiple threads - there are already more than a few that I've listed above. I think the fact that Microsoft Office 12 is going to be multi-threaded is as good a reason as any to show that mathematical applications (like Excel) can be threaded. Now, I also understand that not everything can be multi-threaded for obvious reasons but there are still ways to improve things. Things like Intel's Smart Cache Technology allows one core to steal all of the L2 cache should it need to, meaning that the bandwidth limitations of the memory bus can be reduced even further.
I have used plenty...I am typing this from a Pentium D 930 on XP pro. Professionally I engineer Windows Server solutions for a large corporation.
I know there are applications that use multiple CPU's/Cores. ID's game engines have since Quake 3....although it was buggy. Hence Quake 4 and COD2 on your list. Falcon 4 a popular flight sim does as well. That said most games are GPU starved since DX9 and dont need a lot of CPU except for Flight Sims and RTS games.
I also know that Windows XP/2003 will spread the load of threads (from a non SMP applications) over multiple CPUs/cores. That said it only helps somewhat .compared to say a true SMP aware application like SQL or Exchange where you see a dramatic difference.
Only on a benchmark will you MAYBE see a difference on an equally clocked (same system configuration) single core and dual core box while running a NON SMP aware application.
So until 50% or more applications take advantage of even dual core, multi-core CPUs basically wasted.
Intel and AMD cant ramp the clock speeds faster right now so the new gimmick is to go wide .and so the consumer will be forced into multi core because that is all they will make in a year or so .long before the consumer will see any true benefit from multi core CPUs.
Originally Posted by aggies11 To be fair, while I don't share the voracity of the previous poster( multi-core is definitely NOT a complete waste of time), I feel I have to point out a few things. That all the above apps fall into the category of "media encoding". And there are tons of overlap/redundancy on that list. Yes media encoding is the perfect candidate for parallelization, but it's only one small section of all the types of software that can/is run on a PC.
Office isn't a media encoding programme, and I don't think I would consider WinRAR or Photoshop as 'media encoding' either. Yes, I get your point, media encoding is one of the biggest areas to benefit from parallelism but it isn't the only area. :)
Things like Email and Web browsing are never going to become threaded IMO, but in a world where time is money - the more available threads I've got to use, the more I can get done in a day that already doesn't have enough hours. ;)
I know there are applications that use multiple CPU's/Cores. ID's game engines have since Quake 3....although it was buggy. Hence Quake 4 and COD2 on your list. Falcon 4 a popular flight sim does as well. That said most games are GPU starved since DX9 and dont need a lot of CPU except for Flight Sims and RTS games.
As far as I am aware, Call Of Duty 2 uses its own proprietry engine and it is certainly not OpenGL based - this means that it cannot be based on the Doom 3 engine.
You came across as saying that there are next to no applications that utilise multiple cores, and I was stating that there are an increasing number. Not everything is going to be multi-threaded, but more and more things will be.
Quote:
Only on a benchmark will you MAYBE see a difference on an equally clocked (same system configuration) single core and dual core box while running a NON SMP aware application.
So until 50% or more applications take advantage of even dual core, multi-core CPUs basically wasted.
I disagree - multi-core has improved my own productivity because I am no longer waiting for my CPU to be idle before being able to start something else. I'm a heavy multi-tasker with very little time to get done what needs to be done, so I often find myself doing more than one thing at once. With multi-core, I can work without the need to worry about my operating system slowing down - there are ways to saturate a dual core processor and memory bus and I've found myself doing that on occasion. Multiple cores brings back memories of running a pair of Celerons in an Abit BP6 - things were less-parallel back then, but the smoothness was something that you couldn't get until HyperThreading was announced.
The fact that Conroe is so fast in single-threaded applications as well as multi-threaded applications makes me think that there is potential for even more parallelisation. Four cores with each pair of cores sharing a 4MB L2 cache, means that it should be possible to improve multi-threaded performance too.
Quote:
Intel and AMD cant ramp the clock speeds faster right now so the new gimmick is to go wide .and so the consumer will be forced into multi core because that is all they will make in a year or so .long before the consumer will see any true benefit from multi core CPUs.
If you're talking about Joe Average who uses his computer to talk to friends on Yahoo! Messenger, check their Y! Mail and do very little else.. then no, they're not going to see a benefit because they're not heavy computer users. However, with consoles moving to multi-cores and developers following suit, there are going to be benefits. Both NVIDIA and ATI offload some calculations to a second core with their display drivers.
Of course, very few of the first games on Xbox 360 were multi-threaded because they were ports from PC or alternatively they were developed on a tight time scale. But Microsoft's SDK is supposed to be incredibly good, meaning that developers are going to be able to find ways to use the additional threads available. This can mean only good things for PC gaming. DirectX 10 is another thing that is going to allow developers to make better games, because the specifications are much tighter than they ever have been.
The age of physics is also approaching too. Developers like Crytek decided against using the Havok and Ageia SDKs in favour of developing their own physics engine that uses CPU cycles. I don't think that battle is decided yet either, because it is too early and there are no standards.
There are many ways to use multi-core processors if you're a heavy computer user, but Joe Average (who typically doesn't read bit-tech) is not going to see much of a benefit.
Bindibadgi......"Are we talking about software that was made in 1995 or software that doesnt need to be or wont benefit from being multithreaded?"
Read my original post...."If you look at all of the applications that are currently on the market today for your PC, 95% of them are 32bit and support ONE CORE".
Translation....of all of the software you can purchase right now ..or download or whatever 95% of it is 32bit single core enabled.
"Im good for that bet? Whatcha want on it? What are the definitions of most people? Most people on here or who read these types of reviews on the net? Most people who will be buying system from now until next year?"
July 21 2007 all running PC's in the world...or just in the US...will be on 32bit hardware and OS's...running 32bit applications.
"But you wont be buying a 7900GT if you aren't going to game on it." Game or benchmark. You can easily game on a 7900GT today. Go look at the system specs STEAMPOWERED keeps online that shows the hardware for everyone that connects and runs HL2/Source games. A 7900GT is WAY above most the vast majority of game players. Maybe not for people that visit this site but for the vast majority a 7900GT will last three years when you have people running HL2 on 9600s.
Sure, first produced games wont, but that's because they dont need it. That's the same with every single console and every new bit of hardware: the software has to have time to catch up, it wont always be the case. Future iterations will be produced and will need more processing power to do other stuff. There's room for expansion. You can't bring out an Xbox every year.
Thanks for making my point about not needing a 4 core CPU in your PC. I am in sure in year or more games on the 360 will be mostly multi-core. On the PC it will take even longer. If youre a programmer making 360 games you know that EVERYONE has a 3 core CPU. Not true for a PC software maker, you often write for your lowest common denominator and if you have money to burn you will enhance it for the few that have multi-core CPUs.
That's why it now annually grosses more than movies do and games like WoW have 6 million people playing? That the next generation of consoles is going to be the biggest ever and the graphics card evolution rate has quadrupled that of CPUs... because there's no money in it.
Right you are, in 2003 game sales passed movie theater ticket sales. That said so did sales of DVD movies .by almost double. Games sales total (all platforms) out sell movie tickets. Break that game sales out and PC games, WOW or no WOW, have declined each year since 1998 and console sales have gone up. The ratio is something like 6 to 1 . http://en.wikipedia.org/wiki/Video_game or just google PC game sales.
Originally Posted by Tim S As far as I am aware, Call Of Duty 2 uses its own proprietry engine and it is certainly not OpenGL based - this means that it cannot be based on the Doom 3 engine.
I stand corrected...my appologies.
For Call of Duty 2, Infinity Ward has ditched the Quake 3 engine in favor of all-new proprietary technology. In many regards, it still looks a lot like Call of Duty - the interface looks the same, and the characters look and move in a familiar style. There are some new bells and whistles, however, from the real-time lighting and shadows to the advanced smoke effects that look like they'll play a big part in the game. At first glance, it doesn't pack the "wow" punch of a Half-Life 2 or DOOM 3, but it looks to be a big improvement from the first game.
Comments 26 to 50 of 99
I'd normally agree with you here, but by chance, luck or a weird planet aligment, the first quarter (not fiscally speaking) of 2007 is going to be one of the best moments to upgrade ever. Waiting for 6 months or so will give you acces to Quad-core CPUs, DX10 GPUs and a new OS. When was the last time something similar happened?
Likely, but what if the main benefit is a price drop in Conroe? Waiting a couple or 3 months for Conroe is a good advice anyway, if only so you get a good selection of motherboards (with perhaps more stable BIOS revisions) and new coolers to choose from. Even if I had to upgrade now, I'd wait until I see how nForce5 or RD600 based boards perform or what budget boards will be available. And IIRC, there's at least one new Conroe stepping coming.
So by the time the entire plataform is mature enough and availability and prices have stabilized, Kentsfield will be really near. I agree that the performance jump may not be huge compared with Conroe, but surely is much more sensible to wait a bit and have more options.
Anyone know the max CPU/cores allowed on XP?
A Windows XP license allows you to run two 'processors', but those processors can have an unlimited number of threads (or cores). You can read more here: http://www.microsoft.com/licensing/highlights/multicore.mspx
4 cores should be noticeably fast by the time they are released. People now are beginning to write multi-threaded versions of their applications. In theory if they can write dual core compatible software it shouldn't be much harder to create 4 or 8 core compatible software. It's going from 1 thread to 2 threads which is the hard bit but from there, depending on the software, increasing the number of threads gets easier. We should see lots more massively parallel applications in the future.
Tbh I think that anyone waiting for Quad Core as a desktop platform would be making a big mistake.
Quad Core like Dual core is designed specifically for the server market. There are still only a few applications which are programmed to take advantage of Dual Core processors. The issue is that it cost a whole lot more to develop applications to run on Dual core processors, so just think about the cost for programming Quad core.
For normal users doing a little video encoding, gaming and surfing then Dual Core will be fine. Conroe will continue to evolve in its Dual Core incarnation and owning one will probably be more than enough.
Someone spoke the other day about the fact that were still on 32 bit technology despite so many people owning 64 bit processors. It's unlikley that we will see a complete migration of all applications to 64 bit within foreseable future. As for Quad Core as a viable desktop processor...forget about it. Fully 64 bit & Quad core are currently only viable in the Enterprise Server market, and is outside the realms of reality for 99% of the Computer Enthusiat market.
I agree with the point made about coders not writing for quad core until a good percentage of the market is using machines based on these processors, and the same can be said for the jump from 32 to 64 bit procs.
Personally, my next upgrade around xmas will be stepping up to a AM2 mobo and a AMD X2 chip... mainly because I've been running my machine in it's current config (3200+ Barton, 1GB DDR400, 9600XT) for going on 2 years and it has served me well even up to today (Go AMD!). I get nice frame rates and good visuals even with Elder Scrolls 4, though I will admit I run it at a lower resolution than I'd prefer in order to keep draw distances higher... the lack of HDR support on my gpu doesn't bother me, as I am planning an upgrade in the near future.
In essense, my point is that I have an outdated machine by most of your opinions, but it does everything a top-end rig will do almost as well. I see no practical purpose for an average user to step up to quad-core 64-bit chips, other than bragging rights for having a bleeding-edge rig.
That being said, as software devs make the change as the marketplace does, 6-12 months from now it will probably be a different story.
BTW, anybody else seeing Moore's Law packing it's bags?
It depends on the specific task. Many, if not most, applications arent' very parallelizable. It may take a great deal of effort/difficulty to split that task into only two evenly balanced workloads. Trying to go past that may be either next to impossible, or at least exponentially more difficult.
Certain tasks, for example image rendering, are extremely paralellizable, you can easily divy up the image into 8chunks, because what's in one part does not depend on the other. Few tasks are that accomodating though, as most have tons of linear dependencies and can't be split up that well.
It's not so much a matter of time, it's a matter of difficulty, and often impossibility.
Quad Core on the desktop gains will mostly be seen by the "System Idle Process" ;)
Aggies
Quad-core is unlikely to make a big impact on the market when it first launches in exactly the same way that dual-core didn't, relatively speaking. However, since then, many apps have become multi-threaded - just about every test in our CPU benchmarks suite is multithreaded in one way or another.
The main point I made was slightly hidden, but basically I stand by my original opinion when I reviewed Core 2 Duo. There is little need to spend anymore than US$316 on a Core 2 Duo, because with that you're going to get flagship performance - at least, if you look at AMD's product stack after the price drops come into effect. Buying a flagship Core 2 Duo X6800 is just like buying an Athlon 64 FX-57 - there is no doubting that it is the fastest CPU out there at the moment, but it is soon to be replaced by something with more cores.
In the long run, those cores are going to get used - it's not taken long for developers to develop multithreaded code for dual core processors, so I don't think it will be much more than 12 months before we see massively-multithreaded applications on the market.
Yes I guess you are right that some applications won't be able to fully utilise all the cores. I was thinking more of those applications that have been adapted to run with two cores - they are likely to run well with 4 or more cores.
But then when it comes to performance demanding applications such as games I'm sure we'll see that as developers gain access to more and more power they will find stuff to do to make use of it. For example bolting on more physics and AI where the power to do these tasks is available.
I've read several reports that the introduction of conroe isn't going to be sudden, more a gradual phasing in.
Also, there have been reports that a large portion of the first chips off the line will be going to the likes of DELL and APPLE.
Given the excitement over conroe and the fact that EVERYBODY wants to get one, and a LOT of people have been putting off upgrading till now, is it not possible that INTEL is slightly worried that with limited availability, and the massive price cuts by AMD, some people may be put off from waiting longer and plump for a nice cheap dual core athlon64 ready for the christmas holidays.
By getting everyone excited about something even better that's just around the corner, they can ensure more people will be persuaded to hold off on that upgrade, something they're desparate to do, until kentsfield arrives. By the time people realise that kentsfield will more than likely be out of the price range of many, production should have been able to catch up with the demand for conroe.
This is pure speculation, but there has to be a point to all of this, and INTEL still want to be able to cash in on the Christmas shoppers. If not enough conroes come out, surely, some buyers will be tempted by AMDs low prices.
8-ball
Ad to that intels implied strategy of no longer pushing the frequency envelope and instead pushing the thread count (8 core, and i read an article projecting that intel plans to have a 32 core processor available sometime circa 2010) we could very well see a progressive decrease in the turnaround time it takes software developers to include code which would take advantage of these chips.
The transition from single core to dual core was a unique development. Going from 2x core to 4x core is essentially the same thing.. so it only makes sense that the whole process takes less time. Gonig from 4x to 8x will see additional decrease in turn around time.. etc etc etc. I'm sure you see my point.
Also, another point. With physics and what not in games.. I'm sure games could put the extra cores to work. Granted, it wouldnt be as solid a solution as having a PPU such as Agieas offering.. as they handle data and tasks on completely different levels but 2x cores with physics computations being effectively thrown at it while another two deal with standard cpu functions and background tasks would hardly be something to scoff at.
I was planning on waiting until december and G80 to purchase a new cpu anyway.. as my 805D @ 3.7ghz does alright for me for the time being. This makes me feel good about my decision. What people SHOULD do is always a decision that is heavily effected by perspective and circumstance among many other factors, but waiting for kentsfield certainly is up there with one of the smarter of decisions for those who have upgrade cycles in the 1.5+ year range.
I hate to burst your bubble but if all of those applications are not built to use 4 cores....then most will be fighting for the first one while the other three will be idle.
I got linked to this review/news article. Its sad to see the disinformation that gets put out on the internet.
Some facts really quick
If you look at all of the applications that are currently on the market today for your PC, 95% of them are 32bit and support ONE CORE.
We have had 64bit hardware for over two years now and still the vast majority of applications are 32bit.
Even on platforms such as the 360 most games are NOT using more than a single core.
Waiting for a 4 core CPU to upgrade your personal computer is a total and utter joke.
I would wager than 1 year from now most people will still be running a 32bit operating system with a single core CPU and single threaded applications.
I will go so far as to say that a year from now anyone running Vista on a Conroe will still be running 70-80% of their applications that only support a single core and NONE of them will support more than 2 cores.
If you upgraded your rig to a Conroe and say a 7900GT with 2gigs of ram that it would last most people from 3-5 years. The PC gaming market is shrinking has been for the last 5 years. The only upgrade you might need in those 3 years would be a DX-10 video card if some killer game came out that did not support DX9 backward compatibility .because they only wanted a few sales.
There is so much bad advice on the Internet today.
Games like Elder Scrolls IV: Oblivion, Quake 4, Call Of Duty 2, F.E.A.R., City Of Villians/Heroes and (I think) Prey are multi-threaded and more multi-threaded games are turning up.
These are just a few popular multi-threaded applications that I've listed off the top of my head.
Anyway, welcome to the forums - I hope that you enjoy your stay.
I'm gonna agree with everything you said, although a bit mildier ;)
The fact of the matter, switching to "multi-threaded development" is not simply just a paradigm shift in software development. It's not just "another way of doing things, that given more time you will get better with". There are some (many) tasks that will *never* be multithreaded. Not because programmers are lazy, or it's "new", because it's fundamentally impossible. Some tasks *can't* be broken up into multiple subtasks that can be executed at the same time. Some (many) things are just linear, and unless you can find a way to violate the laws of temporal physics (quantum computing anyone? ;) ), there is no way around it.
Multi-core/parallelization is not the Holy Grail of computing, it is not even "the next logical step". It exists for one reason only, COST. It is simply "cheaper to make 4 slower cores, than one single faster one". The cost and difficulty of developing faster single core chips has risen alot. It's just plain cheaper to stick multiple slower/"easier" cores in that same space. It isn't a perfect solution, because essentially hardware is "passing the buck" onto software.
And quite often, there is nothing software can do about it.
If anything, these facts are the reason why there was so much big fanfare about the rumoured "anti-hyperthreading" or "reverse-hyperthreading" from AMD. Because many things are gonna stay singlecore, and it'd be great to have multi-core "somehow" pretend to be singlecore. But that rumour was obviously wishfull thinking. It can't exist, because it's impossible. There are some problems there are just no ways around. To use a horrible/trivial example: if I want to do the math problem: (4 x 6) + 7. There is no way to break that into two steps that can be done at the same time. No matter how many cores you have, you have to wait for the result of "4 x 6", before you can add 7. These are fundamental laws of space time ;). Many aspects of computing are like that. You can't do the next step, until the results of the previous one are known. That means there is no way to parallelize it.
So yeah, don't get me wrong, if the choice is between 4cores and 2cores for roughly the same price, go for it. It can't hurt. But there is no sense waiting for something that won't see benefits for a long time, if at all. Conroe is real, and is much better. The gains of quad-core are largely theorhetical at this point, and could conceivably be for quite some time.
Remember, 2x2GHz (dualcore) != 1x4GHz. (singlecore)
But 2x2GHz is LOADS cheaper to develop/make than 1x4GHz
Aggies
95% of software is 32bit, but anyone can make up statistics, Kent, 64% of people know that.
Are we talking about software that was made in 1995 or software that doesnt need to be or wont benefit from being multithreaded? Having more cores enables people to do more at once, in the background, having a seemless, smoother computing experience. It's not the end answer by any means, but you it gives developers more to play with and people more to do as more of our lives become embedded in our computers. So with 4 cores you can now download, play a game, listen to music, transcode a DVD and burn another all at the same time. Should you want to get more things done in less time.
USB and PCI took years to take hold. PCI Express has also taken a while. The next Vista will be avaliable in both and drivers are being made avalialbe for both. It's a chicken and egg situation but new CPUs support both, so if something benefits from being 64bit and you're running a 64bit OS, then you get things done faster.
Sure, first produced games wont, but that's because they dont need it. That's the same with every single console and every new bit of hardware: the software has to have time to catch up, it wont always be the case. Future iterations will be produced and will need more processing power to do other stuff. There's room for expansion. You can't bring out an Xbox every year.
Im good for that bet? Whatcha want on it? ;) What are the definitions of most people? Most people on here or who read these types of reviews on the net? Most people who will be buying system from now until next year?
Vista is being made multithreaded. Depends what they use and how many of those programs are running at once. Multithreading is distributed, it's not necessarily designed for a certain core quantity, depending on what it does.
2-3 years, 6-18 months for people on here. But you wont be buying a 7900GT if you aren't going to game on it. If you're just gonna surf the net then buy a mac mini. As larger TFTs come down in price you're gonna need to generate more pixels to fill that screen. I wonder how many people on here are running 3-5 year old graphics cards? 3 years is.. Geforce 4 range, 5 is.. Geforce 256? 2 GTS?
To be fair, while I don't share the voracity of the previous poster( multi-core is definitely NOT a complete waste of time), I feel I have to point out a few things. That all the above apps fall into the category of "media encoding". And there are tons of overlap/redundancy on that list. Yes media encoding is the perfect candidate for parallelization, but it's only one small section of all the types of software that can/is run on a PC.
Multi-threaded is one thing, but I don't think we can say that the games have a balanced load between the cores. 2x2GHz does not perform in-game (CPU wise) relative to a 1x4GHz. There are benefits mind you, but not nearly in the range to say that the games are taking "full advantage" of both cores.
Again, if you offer me a nth-core chip right now, I'm surely gonna take it. But we shouldn't be under the misguided belief that multi-core gives the same benefits of faster single-core, "once the software devs have enough time to get used to it".
I wouldn't be surprised that if tomorrow, magically, 100% of software became as parallelized as possible, the gains would not be nearly as large as everyone expects (or rather, as we have all been lead to believe).
Aggies
Again, though, it's not necessarily about doing just one thing faster.
My move from single core to dual core allowed me to do much more because I am no longer waiting for CPU-heavy tasks to complete. I can now burn a CD/DVD in the background, encode a DVD/video/audio stream in the background. I can work on high resolution images in photoshop without my PC slowing down. I find that the computing experience is completely different - it's hard to explain how and I am sure that many other dual core users will say the same.
Now, I am not expecting the overall computing experience to massively improve with quad-cores, but I am expecting a lot of applications to move to multiple threads - there are already more than a few that I've listed above. I think the fact that Microsoft Office 12 is going to be multi-threaded is as good a reason as any to show that mathematical applications (like Excel) can be threaded. Now, I also understand that not everything can be multi-threaded for obvious reasons but there are still ways to improve things. Things like Intel's Smart Cache Technology allows one core to steal all of the L2 cache should it need to, meaning that the bandwidth limitations of the memory bus can be reduced even further.
I know there are applications that use multiple CPU's/Cores. ID's game engines have since Quake 3....although it was buggy. Hence Quake 4 and COD2 on your list. Falcon 4 a popular flight sim does as well. That said most games are GPU starved since DX9 and dont need a lot of CPU except for Flight Sims and RTS games.
I also know that Windows XP/2003 will spread the load of threads (from a non SMP applications) over multiple CPUs/cores. That said it only helps somewhat .compared to say a true SMP aware application like SQL or Exchange where you see a dramatic difference.
Only on a benchmark will you MAYBE see a difference on an equally clocked (same system configuration) single core and dual core box while running a NON SMP aware application.
So until 50% or more applications take advantage of even dual core, multi-core CPUs basically wasted.
Intel and AMD cant ramp the clock speeds faster right now so the new gimmick is to go wide .and so the consumer will be forced into multi core because that is all they will make in a year or so .long before the consumer will see any true benefit from multi core CPUs.
Things like Email and Web browsing are never going to become threaded IMO, but in a world where time is money - the more available threads I've got to use, the more I can get done in a day that already doesn't have enough hours. ;)
You came across as saying that there are next to no applications that utilise multiple cores, and I was stating that there are an increasing number. Not everything is going to be multi-threaded, but more and more things will be.
The fact that Conroe is so fast in single-threaded applications as well as multi-threaded applications makes me think that there is potential for even more parallelisation. Four cores with each pair of cores sharing a 4MB L2 cache, means that it should be possible to improve multi-threaded performance too.
Of course, very few of the first games on Xbox 360 were multi-threaded because they were ports from PC or alternatively they were developed on a tight time scale. But Microsoft's SDK is supposed to be incredibly good, meaning that developers are going to be able to find ways to use the additional threads available. This can mean only good things for PC gaming. DirectX 10 is another thing that is going to allow developers to make better games, because the specifications are much tighter than they ever have been.
The age of physics is also approaching too. Developers like Crytek decided against using the Havok and Ageia SDKs in favour of developing their own physics engine that uses CPU cycles. I don't think that battle is decided yet either, because it is too early and there are no standards.
There are many ways to use multi-core processors if you're a heavy computer user, but Joe Average (who typically doesn't read bit-tech) is not going to see much of a benefit.
Read my original post...."If you look at all of the applications that are currently on the market today for your PC, 95% of them are 32bit and support ONE CORE".
Translation....of all of the software you can purchase right now ..or download or whatever 95% of it is 32bit single core enabled.
"Im good for that bet? Whatcha want on it? What are the definitions of most people? Most people on here or who read these types of reviews on the net? Most people who will be buying system from now until next year?"
July 21 2007 all running PC's in the world...or just in the US...will be on 32bit hardware and OS's...running 32bit applications.
"But you wont be buying a 7900GT if you aren't going to game on it." Game or benchmark. You can easily game on a 7900GT today. Go look at the system specs STEAMPOWERED keeps online that shows the hardware for everyone that connects and runs HL2/Source games. A 7900GT is WAY above most the vast majority of game players. Maybe not for people that visit this site but for the vast majority a 7900GT will last three years when you have people running HL2 on 9600s.
Sure, first produced games wont, but that's because they dont need it. That's the same with every single console and every new bit of hardware: the software has to have time to catch up, it wont always be the case. Future iterations will be produced and will need more processing power to do other stuff. There's room for expansion. You can't bring out an Xbox every year.
Thanks for making my point about not needing a 4 core CPU in your PC. I am in sure in year or more games on the 360 will be mostly multi-core. On the PC it will take even longer. If youre a programmer making 360 games you know that EVERYONE has a 3 core CPU. Not true for a PC software maker, you often write for your lowest common denominator and if you have money to burn you will enhance it for the few that have multi-core CPUs.
That's why it now annually grosses more than movies do and games like WoW have 6 million people playing? That the next generation of consoles is going to be the biggest ever and the graphics card evolution rate has quadrupled that of CPUs... because there's no money in it.
Right you are, in 2003 game sales passed movie theater ticket sales. That said so did sales of DVD movies .by almost double. Games sales total (all platforms) out sell movie tickets. Break that game sales out and PC games, WOW or no WOW, have declined each year since 1998 and console sales have gone up. The ratio is something like 6 to 1 . http://en.wikipedia.org/wiki/Video_game or just google PC game sales.
I stand corrected...my appologies.
For Call of Duty 2, Infinity Ward has ditched the Quake 3 engine in favor of all-new proprietary technology. In many regards, it still looks a lot like Call of Duty - the interface looks the same, and the characters look and move in a familiar style. There are some new bells and whistles, however, from the real-time lighting and shadows to the advanced smoke effects that look like they'll play a big part in the game. At first glance, it doesn't pack the "wow" punch of a Half-Life 2 or DOOM 3, but it looks to be a big improvement from the first game.
http://pc.gamespy.com/pc/call-of-duty-2/605310p1.html