Well, if this would be a previous generation card I would say that it's the best (in its range).
Considering that it performs few framerates better than a Nvidia GTX 260 that is available for 130 GBP, I can only say that this card is a rip off at 200GBP and a "new generation" joke.
Two GTX 206 for the same amount as one 5850 are a torpedo; not mentioning how old as a technical concept they are.
Conclusion? ATi is apparently learning from AMD.
A 5870x2 with a normal price is the only thing to look forward, if I leave out the prays for the GT300.
That's the problem with most "new generation" hardware, it's only really worth it if you didn't upgrade last generation. I picked up a 4770 for £75, and in all honesty I can't imagine I'll find a game I can't play on my 1440x900 with most of the settings on high for a good few years.
It's worth it if you have a 7900, but if you're sporting a 4850, it's just not worth the investment.
Wait a month or two, prices will drop and most likely there will be a new driver. No need to cut your arms just because new hardware is more expensive than the old stuff....
I bought a Nvidia 260 ONLY as a bridge; expecting the new cards.
My monitor is the HPLp2475w and it needs a little more power.
The only thing I can do with todays ATI's new DX11 cards is to clean my freezer with them; however my freezer has the self cleaning option, so even this does not come as an option (if I don't consider the price of such handywork).
The news just got out that Nvidia is planning to release the mainstream cards of the new Fermi architecture in March next year (if all goes well).
Nvidia is now focusing on the release of the performance parts as a single GT300 and dual "395" version; single should come out as a 230w part.
It seems that the next 5 months won't be very nice to our wallets/purses.
Originally Posted by Kyocera It seems that the next 5 months won't be very nice to our wallets/purses.
Not necessarily. I bet around december the first price cuts will be coming out for these cards. Possibly as low as $200 for a 5850. I just really hope that nvidia has something big up their sleeve with fermi that can drive prices lower
I wonder how the noise level of the HD5850 is. It's quite a bit hotter than the HD5870 in some respects. I'm hoping that part of the reason is a slower-moving and thus quieter cooler.
Originally Posted by azrael- I wonder how the noise level of the HD5850 is. It's quite a bit hotter than the HD5870 in some respects. I'm hoping that part of the reason is a slower-moving and thus quieter cooler.
+1
Indeed, it would be nice to have the noise / sound level of the GPU's included as part of the review as this does affect my own personal buying decision.
Hmm... my 9800GTX died two weeks ago and I don't know what to upgrade with ... I was hoping the HD5850 but
I can get a HD4890 for 140...or the HD5850 for 240... For 100 difference, a bit more Grafphical power, less power requirements, less temp, a quieter fan and DX11...Is it really worth it?
Reading these posts and this review the interesting thing is perspective. My 2.75 year old gtx8800 died two weeks ago and last night I installed my new HD5850 replacement. At £200 I'm getting a ~200% performance increase, a more compact card that runs cooler, uses less power (an important issue with my 520w PSU), is quieter AND costs less than the card it replaced! So, from my perspective this represents an excellent purchase.
I think the DX11 aspect is a bit irrelevant as I bet just like with DX10, it will add overhead that doesn't do much and slows things down. Developers always go for features - not speed (imo).
Indeed, it would be nice to have the noise / sound level of the GPU's included as part of the review as this does affect my own personal buying decision.
What he said. I don't mind a certain amount of noise but could do without a small turbine in my PC.
Any ideas on what triple screen performance is like yet? This is one of the main things that interests me about both this and the 5870 but I'd be interested to see how much of a hammering the frame rate takes from running at 5,040 x 1,050. Do the drivers currently support the Eyefinity features or is this still in the works?
As we've said many times over - it's impossible to accurately record noise levels without sending every product off to a proper sound chamber. We are not happy to publish figures that are scientifically innaccurate without a consistent background noise - our labs are a very busy place, so we prefer to just talk about noise subjectively when things quieten down in there, noting any specific problems in the process. The 5850 is no different to the 5870 in terms of noise from our perspective. Fan pitch stays consistent and it's not loud.
We could, of course, send every product we get to the sound chamber we used for our fan labs, but it adds a significant overhead to a regular review and the result would be delayed reviews. There's very little extra benefit in our opinion - it's not something that can be done in time for most product launches and, frankly, is it worth delaying a review by several days (or up to a week) just to get an accurate noise measurement and add a couple of extra graphs into the review? We don't think it is when everything is considered.
Please don't confuse that with what we'd like to do - of course, we'd love to add scientifically accurate noise performance into all of our reviews, but it's just not realistic without product lead times (ahead of launches) being extended by at least a week (ideally giving us two or three weeks to get everything done). That's unlikely to happen with graphics cards in particular, because the driver team likes to work until the last possible minute to ensure the drivers are as good as possible for the launch reviews.
For example, we had 5870 hardware a week before the drivers arrived and we had less than a week to get all of our 5870 testing done ahead of the launch. Everything, including the testing for the architecture piece was done in that week (the architecture piece was delayed because I spent the few days in the US before IDF in bed with fever - the time I'd set aside to get everything about the 5870 written up). That's pretty damn tight for a brand new architecture as it is because there's generally quite a bit to be explored ahead of a major architecture launch.
Why is the power consumption on idle slightly higher than the HD5870's :|?
OT question: wasn't Direct3D 11 also supposed to run on current D3D10 cards?
Originally Posted by McVittees Reading these posts and this review the interesting thing is perspective. My 2.75 year old gtx8800 died two weeks ago and last night I installed my new HD5850 replacement. At £200 I'm getting a ~200% performance increase, a more compact card that runs cooler, uses less power (an important issue with my 520w PSU), is quieter AND costs less than the card it replaced! So, from my perspective this represents an excellent purchase.
I think the DX11 aspect is a bit irrelevant as I bet just like with DX10, it will add overhead that doesn't do much and slows things down. Developers always go for features - not speed (imo).
Thanks! This is exactly the answer I was looking for!
Regarding the noise testing, I personally don't need accurate decibel measures but simple comparisons to other cards and/or if the noise is a problem at full load. Which is a point that is made on almost every HD4890 review for instance
Nice starting price but I think ill miss this round till more DX11 flood the market. My GFX card is still under 1year old and can manage everything well. Quick question, whats causing the results to even out at the top? is it the card has reached its limit or CPU bottlenecking?
Good review... Im building myself a new comp at the end of the month and i think this is the card im gonna buy. It just about represents best bangperbuck in a current market thats being milked by a dominant ati. Damn nvidea for providing no real competition. I cant justify £300 for the 5870.... I can see the 5850s price dropping to around £170 at some point next year so i wouldnt buy it now if i could wait...but i cant wait :/
Still not 100% sure though...gah! choosing the best value/future proof-ish graphics card atm is hard even with all the great guides on this site.
The earlier recommended route of buying a 4890 for £70 less and waiting for dust to settle/300 series is tempting but i dont see nvidea putting up much of a fight in the price to performance stakes.
Originally Posted by McVittees Reading these posts and this review the interesting thing is perspective. My 2.75 year old gtx8800 died two weeks ago and last night I installed my new HD5850 replacement. At £200 I'm getting a ~200% performance increase, a more compact card that runs cooler, uses less power (an important issue with my 520w PSU), is quieter AND costs less than the card it replaced! So, from my perspective this represents an excellent purchase.
I think the DX11 aspect is a bit irrelevant as I bet just like with DX10, it will add overhead that doesn't do much and slows things down. Developers always go for features - not speed (imo).
Thanks! This is exactly the answer I was looking for!
Regarding the noise testing, I personally don't need accurate decibel measures but simple comparisons to other cards and/or if the noise is a problem at full load. Which is a point that is made on almost every HD4890 review for instance
I've just spoken to James and he said the last time he tried doing some noise testing in house, the equipment he was using picked up the rumble from the Northern Line tube trains heading between Goodge Street and Warren Street every minute or two (the line is about 200-300m from the office). Makes it kinda hard to get even a constant background for recording noise performance when the labs are quiet... short of spending the night in the labs to record noise when the tube is closed.
I guess that's one downside of having our office in Central London.
I've just spoken to James and he said the last time he tried doing some noise testing in house, the equipment he was using picked up the rumble from the Northern Line tube trains heading between Goodge Street and Warren Street every minute or two (the line is about 200-300m from the office). Makes it kinda hard to get even a constant background for recording noise performance when the labs are quiet... short of spending the night in the labs to record noise when the tube is closed.
I guess that's one downside of having our office in Central London.
Double Glazing? :oP
Fair comments on the more precise /scientific style noise level testing but a mention as to whether a particular card seems overly loud or particularly quiet would be of some use and thanks for clarifying that this card is does not seem to be loud. Thanks Tim ;o)
I guess measuring noise levels is something the manufacturers should realistically be doing.
Nice review, although i'm enjoying my 5870. Can't believe the difference in noise/heat compared to the 4850's and 4870's i owned previously. Means i actually have to turn the central heating on now though, rather than rely on my PC hardware to heat the room..
I feel the 5850 worth the price tag. Its essentially two cards from the last generation in one while staying at the same price it would cost for the older generation cards. Of coarse, its more energy efficient and supports DX11. I may purchase this over the 5870 for the price/performance factor.
Comments 1 to 25 of 34
ReplyConsidering that it performs few framerates better than a Nvidia GTX 260 that is available for 130 GBP, I can only say that this card is a rip off at 200GBP and a "new generation" joke.
Two GTX 206 for the same amount as one 5850 are a torpedo; not mentioning how old as a technical concept they are.
Conclusion? ATi is apparently learning from AMD.
A 5870x2 with a normal price is the only thing to look forward, if I leave out the prays for the GT300.
It's worth it if you have a 7900, but if you're sporting a 4850, it's just not worth the investment.
My monitor is the HPLp2475w and it needs a little more power.
The only thing I can do with todays ATI's new DX11 cards is to clean my freezer with them; however my freezer has the self cleaning option, so even this does not come as an option (if I don't consider the price of such handywork).
The news just got out that Nvidia is planning to release the mainstream cards of the new Fermi architecture in March next year (if all goes well).
Nvidia is now focusing on the release of the performance parts as a single GT300 and dual "395" version; single should come out as a 230w part.
It seems that the next 5 months won't be very nice to our wallets/purses.
Not necessarily. I bet around december the first price cuts will be coming out for these cards. Possibly as low as $200 for a 5850. I just really hope that nvidia has something big up their sleeve with fermi that can drive prices lower
+1
Indeed, it would be nice to have the noise / sound level of the GPU's included as part of the review as this does affect my own personal buying decision.
I can get a HD4890 for 140...or the HD5850 for 240... For 100 difference, a bit more Grafphical power, less power requirements, less temp, a quieter fan and DX11...Is it really worth it?
I think the DX11 aspect is a bit irrelevant as I bet just like with DX10, it will add overhead that doesn't do much and slows things down. Developers always go for features - not speed (imo).
What he said. I don't mind a certain amount of noise but could do without a small turbine in my PC.
Any ideas on what triple screen performance is like yet? This is one of the main things that interests me about both this and the 5870 but I'd be interested to see how much of a hammering the frame rate takes from running at 5,040 x 1,050. Do the drivers currently support the Eyefinity features or is this still in the works?
We could, of course, send every product we get to the sound chamber we used for our fan labs, but it adds a significant overhead to a regular review and the result would be delayed reviews. There's very little extra benefit in our opinion - it's not something that can be done in time for most product launches and, frankly, is it worth delaying a review by several days (or up to a week) just to get an accurate noise measurement and add a couple of extra graphs into the review? We don't think it is when everything is considered.
Please don't confuse that with what we'd like to do - of course, we'd love to add scientifically accurate noise performance into all of our reviews, but it's just not realistic without product lead times (ahead of launches) being extended by at least a week (ideally giving us two or three weeks to get everything done). That's unlikely to happen with graphics cards in particular, because the driver team likes to work until the last possible minute to ensure the drivers are as good as possible for the launch reviews.
For example, we had 5870 hardware a week before the drivers arrived and we had less than a week to get all of our 5870 testing done ahead of the launch. Everything, including the testing for the architecture piece was done in that week (the architecture piece was delayed because I spent the few days in the US before IDF in bed with fever - the time I'd set aside to get everything about the 5870 written up). That's pretty damn tight for a brand new architecture as it is because there's generally quite a bit to be explored ahead of a major architecture launch.
OT question: wasn't Direct3D 11 also supposed to run on current D3D10 cards?
Thanks! This is exactly the answer I was looking for!
Regarding the noise testing, I personally don't need accurate decibel measures but simple comparisons to other cards and/or if the noise is a problem at full load. Which is a point that is made on almost every HD4890 review for instance
Run the Crysis benchmark and be happy. :)
Still not 100% sure though...gah! choosing the best value/future proof-ish graphics card atm is hard even with all the great guides on this site.
The earlier recommended route of buying a 4890 for £70 less and waiting for dust to settle/300 series is tempting but i dont see nvidea putting up much of a fight in the price to performance stakes.
I've just spoken to James and he said the last time he tried doing some noise testing in house, the equipment he was using picked up the rumble from the Northern Line tube trains heading between Goodge Street and Warren Street every minute or two (the line is about 200-300m from the office). Makes it kinda hard to get even a constant background for recording noise performance when the labs are quiet... short of spending the night in the labs to record noise when the tube is closed.
I guess that's one downside of having our office in Central London.
for that you need 5870 or duel gpu simple as that
Double Glazing? :oP
Fair comments on the more precise /scientific style noise level testing but a mention as to whether a particular card seems overly loud or particularly quiet would be of some use and thanks for clarifying that this card is does not seem to be loud. Thanks Tim ;o)
I guess measuring noise levels is something the manufacturers should realistically be doing.
-
« Previous
-
1
-
2
-
Next »
Discuss in the forums