Hi all, Gareth here from Corsair to answer a few questions.
The H50 is indeed based on the Asetek LCLC, but we worked with Asetek for some considerable amount of time before we agreed on the final design.
There have been many revisions of the LCLC since it launched, primarily focusing on the pump / cold plate unit. The Corsair H50 uses a new version of the pump / cold plate, and this is what gives it the extra performance. Older revisions of the LCLC do not offer nearly the same performance, and we tested several recent revisions.
Here's an image of the cold plate of an older LCLC (in this case a NorthQ) and the H50. As you can see, the cold plate design is quite different.
These are some comparative results from a test system I have set up here. The chassis was a fairly compact midi-tower, so overall the numbers for all three will be a bit higher than you'd see in a larger / more typical chassis. The 'LCLC' was set up in the same way as the H50, and I even used the H50's fan and CPU mounting bracket for it.
Adding blocks would be asking for trouble. As might adding a larger rad. No, I think this little baby does what it's meant to and nothing more. Which is fine for 50GBP :)
Originally Posted by Ice Tea Any plans on a bigger Radiator and more Blocks for the NorthBridge or Graphics card ?.
The H50 is a CPU cooler and the aim was to provide a higher-performance alternative to a high-end HSF, for a similar price. Spawning more blocks and adding a larger rad would price it out of that segment, and put it more into traditional water-cooling territory.
We are obviously looking at ways to expand our range of cooling products though.
I will have mine in a couple of days and can send you a PM with the figures if you want?
That would be great, I was looking at ordering one in about a week. Let me know how it works out for you and how you ended up mounting it (what case you have?).
Also on a side note, having looked at the comparison photo's you can see in the one on the left (the LCLC) why I lapped mine. It is not a smooth surface. The new H50 looks to have a perfectly flat copper surface which is nice.
Originally Posted by B1ackbeard The H50 is a CPU cooler and the aim was to provide a higher-performance alternative to a high-end HSF, for a similar price. Spawning more blocks and adding a larger rad would price it out of that segment, and put it more into traditional water-cooling territory.
We are obviously looking at ways to expand our range of cooling products though.
Quick technical question. Does the orientation of the rad or CPU block matter much? eg, if the rad was mounted on the roof and the block was mounted with the pipes connected at the top. Thanks. :)
I look forwards to a verion for GFX cards. That would arguably be even more usefull!
I haven't tried mounting the CPU block upside down (180 degree rotated) but I can't see how it would make any difference.
I have tried different radiator positions, though. A roof mounting position was pretty much directly comparable to a rear mount, although I did see some benefit from mounting on a side panel (with the rear fan acting as an exhaust), which I put down to the fact that intake air was being blown directly onto the CPU area of the motherboard. It makes fitting/removing the side panel a bit trickier though! :D
It really depends on your chassis, and how you can best set up the airflow.
Originally Posted by Bindibadgi No, the article clearly states that for our "AMD test setup" that includes this particular case and CPU (etc) together, it doesn't work so well. You can't take a single CPU at a single overclocked and overvolted value and apply it to the entire "AMD" line of CPUs anyway, otherwise we might as well test it on a bench outside a case which isn't a real world scenario. If you go look up other reviews and see their AMD numbers, in whatever case they use, it'll give you a wider idea.
If you just look at the charts and pictures you are doing yourself an injustice in any instance, that's what words are written for.
This confuses me even more. How can you say the test shows that for this AMD processor the H50 isn't worthwhile while at the same time admitting that the case airflow is what caused the poor performance. Either you should have tested all in the same case or used the same internals and changed the case. As it stands your tests are meaningless.
They only show that the performance is going to be entirely dependant on your setup, which I don't need a review to tell me. If the low performance was due to the mounting system or because the heat produced exceeded the cooler's ability to dissipate then that would be a very useful piece off information to have. But this testing method doesn't tell me that.
I'm not saying that an exhaustive scientific approach needs to be taken but some consistency would be nice.
PS: I did read the text but that doesn't change the fact that the charts (indeed the whole testing method) are misleading.
i wonder if one could shoehorn this into a qpack2 ... initial thought is that the hoses might not be long enough, or that the fan area would not have enough spare room for the rad ...
Originally Posted by Cobalt This confuses me even more. How can you say the test shows that for this AMD processor the H50 isn't worthwhile while at the same time admitting that the case airflow is what caused the poor performance. Either you should have tested all in the same case or used the same internals and changed the case. As it stands your tests are meaningless.
They only show that the performance is going to be entirely dependant on your setup, which I don't need a review to tell me. If the low performance was due to the mounting system or because the heat produced exceeded the cooler's ability to dissipate then that would be a very useful piece off information to have. But this testing method doesn't tell me that.
I'm not saying that an exhaustive scientific approach needs to be taken but some consistency would be nice.
PS: I did read the text but that doesn't change the fact that the charts (indeed the whole testing method) are misleading.
Quote:
Originally Posted by From the Review The AMD results may seem like a bit on the high side but we tested, and retested, and remounted, and checked and changed the pump and fan connectors, but the results were entirely consistent. The problem is that the Cosmos 1000 lacks any intake fans, it only has two 120mm exhaust fans in the roof and a 120mm exhaust fan in the above the I/O shield. As a result, even when we set the Corsair H50 to draw air in from outside the case the system ran very hot.
It's a worrying trend that pairing a case without positive pressure or a ready route for air to escape through a mesh means performance suffers massively. After the CPU had been under load for an hour the motherboard around the CPU socket was to hot it wasn't comfortable to touch, while the CPU block was also very hot. Worse still, the air temperature inside the case was in excess of 45C. Relatively this makes it worse than the stock AMD cooler - i.e. clearly very, very poor.
Clearly with a normal heatsink the two roof fans draw off the heat right out the case, leading to better performance, but with the radiator on the H50 limiting airflow, it really makes the whole setup struggle. Evidently the H50 isn't suitable for use in every case.
If you read the above that is put underneath the AMD graph, to explain it, not once do I allude it was to the CPU alone.
AMD/Intel CPUs are simply heat generators. No one CPU is representative of an entire socket. If anyone claims to test "AMD" alone is misinforming their readers. The only thing you can test is how easy it is to install on that socket and if the contact works effectively, but other than that, all a heatsink has to do is suck away the heat generated. All our CPUs are over volted in a "typical overclock" but there's no way we could test every CPU, every overclock, every motherboard, etc. What is tested is a typical system: an environment. NOT a CPU.
From your accusation about meanless testing we might as well not test in a case at all or might as well just used a hotplate of CPU surface area size.
If we had only used a Stacker in all our tests we would conclude that the H50 is fantastic and would recommend it for everyone, which is clearly NOT the case. Some people would buy it and get worse temps than they would have with a cheaper CPU cooler because their case is better designed to make use of that instead if it doesn't have many intake fans.
I do concede perhaps we should test cases and CPUs separately but it would not represent a real world environment or system, and it would take an exponential amount of time considering the MANY inevitable requests we would have to review x, y, z case. It's simply infeasible for any review site. Perhaps we should use three entirely different cases representing three completely different builds?
Any graph out of context can direct wrong information and assumption, that's why we add the entire setup list and results discussion to explain our findings.
I have gone through and amended the graphs and page subtitles to be more accurate though, indicating that they are systems not CPU sockets alone. Hopefully this will halt the cause some confusion. :)
Originally Posted by Rocket_Knight64 Quick technical question. Does the orientation of the rad or CPU block matter much? eg, if the rad was mounted on the roof and the block was mounted with the pipes connected at the top. Thanks. :)
I look forwards to a verion for GFX cards. That would arguably be even more usefull!
Wonder if they could add a mount for the current Pump/Block to work on GFX cards ?
I'm more than happy with my modified Swiftech H20-120 but it's nice to see more of these smaller, easy install, kits coming onto the market than actually work.
My Rig is based on a Small Form Factor using a Lian Li PC-V351B case + Asus Rampage II Gene microATX MB + Core i7-920 rev C0 + 3 x 2 GB Corsair Dominator 1600Mhz cl8 DDR3 + WD RE3 1TB + HD4850 fanless GPU
My Rig is so hot I can't barely load it and Forget about OC : NB is the
worth, its temp reaches 78C while CPU temp goes beyond 85C with OCCT
My Noctua C12P is giving up with such temps !
Thus I greatly interested by H50 but what about cooling NB , SB , moffset
Originally Posted by CyrIng My Rig is based on a Small Form Factor using a Lian Li PC-V351B case + Asus Rampage II Gene microATX MB + Core i7-920 rev C0 + 3 x 2 GB Corsair Dominator 1600Mhz cl8 DDR3 + WD RE3 1TB + HD4850 fanless GPU
My Rig is so hot I can't barely load it and Forget about OC : NB is the
worth, its temp reaches 78C while CPU temp goes beyond 85C with OCCT
My Noctua C12P is giving up with such temps !
Thus I greatly interested by H50 but what about cooling NB , SB , moffset
Your problem must be airflow, and the V351B isn't exactly best known for its airflow!
All water cooling dose is get you closer to ambient temp, and the ambient temp around your radiator would have to be room temp, and not the internal case temp. which would be a problem as your running a passively cooled ATI 4850, which is dumping a huge amount of heat in there!
Take your side panels off after an hour of use and feel the air temp inside, it'll be seriously toasty.
Originally Posted by Burnout21 ... the V351B isn't exactly best known for its airflow!
You're right ;) I'm still wondering if I should persue with SFF (that I love so much) or spend energy in more serious cases.
Quote:
Originally Posted by Burnout21 ... Take your side panels off after an hour of use and feel the air temp inside, it'll be seriously toasty.
It is toasty
I connected a MB's temp sensor between the heatsink & the pipes, temp @ idle is 52°C ( GPU core @ 62°C )
While MB temp reports 42-45°C ( during this summer ) , I will take the time to replace with an old video PCI card to measure ambient temp. If it reaches arround 35°C I can go for WC.
I read this review, have just read the custom pc review, have read reviews on line but I have one question.
Were can you buy it in the UK? Who has this in stock? This unit has been on back order since late July, whatst the point of every one raving about this if you cant purchase it.
I am considering going full water or just getting the best air cooler I can.
A limited number of H50s should be arriving in the UK this week at the usual places, such as Scan, OC UK and so on, and then in greater numbers in a couple more weeks.
Sorry for the delay, but rest assured they are coming!
Just fitted my corsair hydro h50-1 to my below spec. it has drastically dropped temps and raised overclocking headroom by, I don't know how far yet!
Install was pretty good. however my case's airflow pretty much limits me to an exhaust, I've got the 1700rpm as the push fan and a scythe 1100rpm fan as the pull. thing is it's a small-ish midi case, with no side panel airholes to speak of, and two 80mm fans at the front. so limited airflow indeed.
however I'm still seeing temps, with my 260 at 80C below it, with a 1.58vcore, 268htt and 14multi of less than 65C full prime. the chip wasn't even stable at that on my previous scythe infinity.
Overall a brilliant product that simply works. and works very very well ^_^
Comments 51 to 75 of 83
ReplyThe H50 is indeed based on the Asetek LCLC, but we worked with Asetek for some considerable amount of time before we agreed on the final design.
There have been many revisions of the LCLC since it launched, primarily focusing on the pump / cold plate unit. The Corsair H50 uses a new version of the pump / cold plate, and this is what gives it the extra performance. Older revisions of the LCLC do not offer nearly the same performance, and we tested several recent revisions.
Here's an image of the cold plate of an older LCLC (in this case a NorthQ) and the H50. As you can see, the cold plate design is quite different.
http://forums.bit-tech.net/picture.php?albumid=408&pictureid=4187
These are some comparative results from a test system I have set up here. The chassis was a fairly compact midi-tower, so overall the numbers for all three will be a bit higher than you'd see in a larger / more typical chassis. The 'LCLC' was set up in the same way as the H50, and I even used the H50's fan and CPU mounting bracket for it.
http://forums.bit-tech.net/picture.php?albumid=408&pictureid=4189
Any plans on a bigger Radiator and more Blocks for the NorthBridge or Graphics card ?
.
The H50 is a CPU cooler and the aim was to provide a higher-performance alternative to a high-end HSF, for a similar price. Spawning more blocks and adding a larger rad would price it out of that segment, and put it more into traditional water-cooling territory.
We are obviously looking at ways to expand our range of cooling products though.
Seriously considering one of these now.
Thanks for the rapid response.
That would be great, I was looking at ordering one in about a week. Let me know how it works out for you and how you ended up mounting it (what case you have?).
Thanks :)
I look forwards to a verion for GFX cards. That would arguably be even more usefull!
I have tried different radiator positions, though. A roof mounting position was pretty much directly comparable to a rear mount, although I did see some benefit from mounting on a side panel (with the rear fan acting as an exhaust), which I put down to the fact that intake air was being blown directly onto the CPU area of the motherboard. It makes fitting/removing the side panel a bit trickier though! :D
It really depends on your chassis, and how you can best set up the airflow.
http://users.telenet.be/OAP2/AMDH50.png
words and explenations in the review...
This confuses me even more. How can you say the test shows that for this AMD processor the H50 isn't worthwhile while at the same time admitting that the case airflow is what caused the poor performance. Either you should have tested all in the same case or used the same internals and changed the case. As it stands your tests are meaningless.
They only show that the performance is going to be entirely dependant on your setup, which I don't need a review to tell me. If the low performance was due to the mounting system or because the heat produced exceeded the cooler's ability to dissipate then that would be a very useful piece off information to have. But this testing method doesn't tell me that.
I'm not saying that an exhaustive scientific approach needs to be taken but some consistency would be nice.
PS: I did read the text but that doesn't change the fact that the charts (indeed the whole testing method) are misleading.
If you read the above that is put underneath the AMD graph, to explain it, not once do I allude it was to the CPU alone.
AMD/Intel CPUs are simply heat generators. No one CPU is representative of an entire socket. If anyone claims to test "AMD" alone is misinforming their readers. The only thing you can test is how easy it is to install on that socket and if the contact works effectively, but other than that, all a heatsink has to do is suck away the heat generated. All our CPUs are over volted in a "typical overclock" but there's no way we could test every CPU, every overclock, every motherboard, etc. What is tested is a typical system: an environment. NOT a CPU.
From your accusation about meanless testing we might as well not test in a case at all or might as well just used a hotplate of CPU surface area size.
If we had only used a Stacker in all our tests we would conclude that the H50 is fantastic and would recommend it for everyone, which is clearly NOT the case. Some people would buy it and get worse temps than they would have with a cheaper CPU cooler because their case is better designed to make use of that instead if it doesn't have many intake fans.
I do concede perhaps we should test cases and CPUs separately but it would not represent a real world environment or system, and it would take an exponential amount of time considering the MANY inevitable requests we would have to review x, y, z case. It's simply infeasible for any review site. Perhaps we should use three entirely different cases representing three completely different builds?
Any graph out of context can direct wrong information and assumption, that's why we add the entire setup list and results discussion to explain our findings.
I have gone through and amended the graphs and page subtitles to be more accurate though, indicating that they are systems not CPU sockets alone. Hopefully this will halt the cause some confusion. :)
If anyone has used it on an ATCS 840 please share results, I'm still thinking about this one.
Wonder if they could add a mount for the current Pump/Block to work on GFX cards ?
My Rig is so hot I can't barely load it and Forget about OC : NB is the
worth, its temp reaches 78C while CPU temp goes beyond 85C with OCCT
My Noctua C12P is giving up with such temps !
Thus I greatly interested by H50 but what about cooling NB , SB , moffset
Your problem must be airflow, and the V351B isn't exactly best known for its airflow!
All water cooling dose is get you closer to ambient temp, and the ambient temp around your radiator would have to be room temp, and not the internal case temp. which would be a problem as your running a passively cooled ATI 4850, which is dumping a huge amount of heat in there!
Take your side panels off after an hour of use and feel the air temp inside, it'll be seriously toasty.
I connected a MB's temp sensor between the heatsink & the pipes, temp @ idle is 52°C ( GPU core @ 62°C )
While MB temp reports 42-45°C ( during this summer ) , I will take the time to replace with an old video PCI card to measure ambient temp. If it reaches arround 35°C I can go for WC.
CyrIng
Were can you buy it in the UK? Who has this in stock? This unit has been on back order since late July, whatst the point of every one raving about this if you cant purchase it.
I am considering going full water or just getting the best air cooler I can.
When can I buy a H50?
Sorry for the delay, but rest assured they are coming!
Install was pretty good. however my case's airflow pretty much limits me to an exhaust, I've got the 1700rpm as the push fan and a scythe 1100rpm fan as the pull. thing is it's a small-ish midi case, with no side panel airholes to speak of, and two 80mm fans at the front. so limited airflow indeed.
however I'm still seeing temps, with my 260 at 80C below it, with a 1.58vcore, 268htt and 14multi of less than 65C full prime. the chip wasn't even stable at that on my previous scythe infinity.
Overall a brilliant product that simply works. and works very very well ^_^
-
« Previous
-
1
-
2
-
3
-
4
-
Next »
Discuss in the forums