Blog post: "Unified GPU/SMP benchmarking scheme ..."

[kiore]

I think you are underestimating the efficiency and simplicity of the 4P systems. ATLASFOLDER made a great contribution, but card fans wear out and are not made to run 24/7/365 at 100%. Could you be running a 4P system for 5 years? Why not, this is not an extreme life for a server. I doubt the average GPU rig will hold up so well.

Well, Ati-HD4xxx was released 25.06.2008 while Folding@home dropped support 01.09.2011 meaning roughly 3 years, so it's unrealistic to expect a GPU will be usable in FAH for 5 years.

Also, GPU-fans isn't normally running at 100%, example my fans is at 35% for the moment meaning not much higher than idle speed, so for the expected crunching-life of a GPU until obsolete wouldn't expect any more problems with GPU-fans than with cpu-fans.

You may live it a cold place and have only 1 GPU. what is being referred to is multi GPU rigs which will normally require max fan speeds, the Atlas folder rigs were 4 double GPUs just millimeters apart. The complexity of running multi GPU systems 24/7 should not be overlooked, it may sound simple just populating vacant pcie slots with new cards but the heat, power supply and airflow factors are not to be underrated.

[mdk777]

example my fans is at 35% for the moment meaning not much higher than idle speed

depends on the card and the wattage. Running high end cards, like the 580 under discussion recently, you are much better off running the fans at higher than stock settings. Cooling not only the GPU, but also the VRM is a consideration for card life in the long haul. Case fans, ambient temps, dust accumulation, are all factors that enter in over time when GPU folding. I'm not saying the fans will die overnight...just that the usage is higher than standard gaming use and stress will have an effect over time. Conversely, SERVER MB and CPU are designed for long life and relatively high continuous use. Everything is relative...just throwing out some other variables other than just price and points.

[Punchy]

The only thing that ever changes the announced benchmark settings is public outcry and I do not think you are going to see that happen here. The group of people who are usually doing that are being very silent this time because this move benefits them (Human Nature)

Funny, if this thread and several others regarding GPU QRB points vs other points aren't public outcry, then I don't know what is. The human nature of which you speak is already in action here and in other forums.

I think most open-minded folks are not commenting on the too high / too low issue until the full complement of data is available - that is, the full details of the SMP project to be compared 1:1 to p8057. Without that data there is no way to make a judgement on the current values.

[tear]

Without that data there is no way to make a judgement on the current values.

Has already been done.

AMD dual Hex @ 2.4
P8055: 1m00s
P8057: 15m27s

4P MC @ 3.0
P8055: 0m17s
P8057: 8m01s

Plus: http://hardforum.com/showthread.php?p=1039277090

This is not about too low / too high. It's about general implications which is what Grandpa's talking about.

[Evil Penguin]

It's no secret that smaller proteins fold faster on GPUs than on CPUs.
Some of you guys are jumping the gun based on one small WU.

Let's see how bigadv WUs do on GPUs.

[tear]

Once explicit/cut-off simulations hit GPU street we can have that conversation. Right now there are none.

EDIT: plus, the only explicit project that was run on GPU has been withdrawn due to instability;
that tells us GPUs will continue to run implicit simulations _exclusively_ for quiiiite some time, heh, heh

[Grandpa_01]

The points appear to be correct using = pay for = work the only thing missing is scientific value which I have no idea if there is such a thing in the point system. If things remain the same there is going to be a very large imbalance in the folding arena, does it matter, only PG knows the answer to that question. All that I know is that if things remain the same my 4P rigs will go off line and be replaced with GPU rigs, I am not griping about this it saves me quite a bit both in power and equipment cost PG determines what I fold with, with the rewards system and that is just the way it is

I am actually happy with this move and hope they can port all WU's over to GPU it will increase the amount of science being done by quite a bit, hopefully that is the case because it is going to be a hard sale to get donors to fold anything else.

[Punchy]

This is not about too low / too high. It's about general implications which is what Grandpa's talking about.

I was merely addressing the OP's innuendos regarding too high / too low and "groups of people" - I didn't bring up the topic.
And, you're fooling yourself if you don't believe that points is inextricably linked to pretty much every argument on this forum.

[tear]

You're reading too much into this.
Grandpa's more than capable of shifting to 16 GTX 580
in no to little time (compared to 4Ps it's a piece of cake).
He'll get the points either way.

Though is that kind of shift (on global scale) what the project really wants?

[csvanefalk]

Though is that kind of shift (on global scale) what the project really wants?

While I am stuck with SMP since I run Linux, I think this is the way computing in general is going. It is very probable that the next decade will see more and more blurring of the boundaries between CPU and GPU, at the very least on the API level. Ever since OpenCL and related libs were introduced, this has only felt natural. I would actually commend PG for putting focus on GPU computing, in the end I think that is what will benefit the actual science the most (even if us poor Linux users will be left hopelessly behind PPD wise until we finally get our own GPU client).

[Grandpa_01]

Though is that kind of shift (on global scale) what the project really wants?

While I am stuck with SMP since I run Linux, I think this is the way computing in general is going. It is very probable that the next decade will see more and more blurring of the boundaries between CPU and GPU, at the very least on the API level. Ever since OpenCL and related libs were introduced, this has only felt natural. I would actually commend PG for putting focus on GPU computing, in the end I think that is what will benefit the actual science the most (even if us poor Linux users will be left hopelessly behind PPD wise until we finally get our own GPU client).

The test that have been done folding GPU (using version 6 of tah with Wine on Linux) has the same results as Windows GPU folding (8057) and I am sure there will be many good guides out soon for it. So Linux users can fold GPU and enjoy the same production.

[csvanefalk]

The test that have been done folding GPU (using version 6 of tah with Wine on Linux) has the same results as Windows GPU folding (8057) and I am sure there will be many good guides out soon for it. So Linux users can fold GPU and enjoy the same production.

I'm aware of the Wine hack, and I am grateful for all the effort that has gone into it. It would be very nice to have a native client, but if that is the best we will get, then so be it...always better than nothing

[mdk777]

I'm not sure they need or want the shift to happen instantly.
I'm also not sure it(the majority of donations) will shift as fast as the elite folders here might.

However, the green 500 list indicates that it currently the best way to get the most compute for the least wattage.

http://www.anandtech.com/show/6457/gree ... amd-nvidia

I'm all for the most efficient use of my donation as possible.

[7im]

Well, Ati-HD4xxx was released 25.06.2008 while Folding@home dropped support 01.09.2011 meaning roughly 3 years, so it's unrealistic to expect a GPU will be usable in FAH for 5 years.

Also, GPU-fans isn't normally running at 100%, example my fans is at 35% for the moment meaning not much higher than idle speed, so for the expected crunching-life of a GPU until obsolete wouldn't expect any more problems with GPU-fans than with cpu-fans.

Seems like a mischaraterization of the whole by cherry picking and misrepresenting facts.

First, a lot of NV cards have been and are still folding much longer than the example cited. History speaks for itself.

Next, FAH did drop support, but not by choice. ATI ended Brook support, so FAH was forced to go to OpenCL, which forced the end of the 4xxx and lower cards. Balance that against OpenCL which will be here a long time. And don't forget CUDA has been around even longer.

As for fan life, you get what you pay for. That $10 rebate on a lowest end card will cost you eventually...

[Leonardo]

Grandpa, Ben Lamb, Tear, and others: Thanks for a good discussion.