The new GTX 460 could be quite good in terms of performance for gpugrid. Possibly faster than a GTX480 with a little overclock.

http://www.anandtech.com/show/3809/nvidias-geforce-gtx-460-the-200-king

GDF

If you have one let us know here, to make some benchmark with the beta tests.

I think you are getting carried away. I expect it will be faster than a GTX285 and a GTX465. It might even come close to a GTX470, especially if you get an 800MHz one, but I doubt it will do more than a GTX480. That said two GTX460's should do more than one GTX480, and cost about the same.



Also,
http://images.anandtech.com/graphs/gtx460_071110174503/23755.png
http://images.anandtech.com/graphs/gtx460_071110174503/23756.png

Interestingly in the above test there was no difference between the cheaper and more expensive versions of the card when it came to crunching at Folding. We will have to wait and see if that holds true here.

GDF, perhaps if you had a fixed test WU to crunch people would be able to bench according to that, and reviewers would add GPUGrid to their reviews!

Maybe we will have to wait for the GTX475 then for a faster performance.
The new design of the cores makes much more sense and it's perfect for our application.
I am still curious to test on GPUGRID. How do they do at F@H to give the benchmark away?

gdf

Hopefully the new design will actually work better for GPUGrid than for Folding@Home, but these cards seem to have been introducing more technology rather than trying to improve on existing performances. That said, I agree the architecture is much better, and the cards are more sellable! I’m slightly worried about the shaders, but we wont know for sure how they perform until someone actually buys one, attaches it to GPUGrid and successfully runs a work unit. I think they will be an excellent card to buy for those just wanting one descent card in their system; they look like good value for money.

It’s probably still best just to look at the GTX460 compared to the GTX465, GTX470 and GTX480, rather than the GT200s. All the Fermi’s are likely to benefit from driver and CUDA improvements relatively evenly in the future, while any such improvements would make all the Fermi cards improve against the GT200 cards. So is the GTX460 better value for money than a GTX465, GTX470 or GTX480, and do they do more work per Watt? With respect to gaming the answer is Yes and Yes, but we have to wait and see about GPUGrid performance.

I think the Folding test just involves folding a known Protein, using different GPU cards and then comparing the time taken, and power used.
So for example, if they folded a reference, 236aa transcription regulator CL Protein (Lambda Repressor), using one app they could test multiple cards relative performances. In six months or a year when a new card comes out, as long as they use the same app and other hardware, then they don’t have to test all the cards again, just the new one!

At the time being Fermi's still only work well on Linux and XP when crunching here - there is a 40% hit running on W7.

That graph style is Anandtech, but I can't find any GP-GPU stuff in the 2 launch day reviews. Where is it from? I thought it was unsatisfying to bench games only, as there's not much to see there anyway..

@GDF: when I ran F@H (a few years ago) it was basically a command line app. For benchmarking purposes people mostly compared run times of the same project (=different clients / algorithms) and same WU (=protein I think, they'd just conquered nr. 2000 back then) which saw small variance.
However, for reliable benchmarking it may have been possible (and could still be) to copy a WU file into the same directory and launch the client with other (or none) parameters. It would then just process the data, write an output file but don't attempt to upload anything.

Actually.. the more I think about it, the less I trust my memory on this one. Could have mixed it up with some other project.

MrS
____________
Scanning for our furry friends since Jan 2002

We have to test on one to know the performance. I am just saying that it is better designed and should allow higher overclock than the original Fermi.

gdf

There is no doubt it is more competitively designed – designed to sell, and they will, partially because they will overclock well; 800MHz for the GPU should be the standard, but the problem might be the memory controller. I suspect this is not the case and that Beyond made a good choice; nice high factory clock, the less expensive 768MB card, and with a lifetime warranty :) I'm guessing at least 30% faster than a GTX260. In the UK we can pick a GTX460 up for just over £150 - My GTX470 cost more than twice that!
Although I expect there will be little or no difference in performance for cards with more RAM (1GB), we will have to wait and see. I hope these do not suffer from driver issues, as is the case with GT240's that only have 512MB (Vista/W7 only).
Also, when the CUDA 3.1 issue is resolved there could be significant improvements.

MrS, I remember something similar.

@SnowCrash:
I'd also go for a factory clocked one, if OC'ed ones cost more and OC myself. The factory OC'ed ones might be binned for better chips - but who knows. The difference is not going to be huge.

The amount of 786 MB RAM should be fine. The main difference is 33% more bandwidth and L2 cache for the 1 GB version (the 33% more ROPs don't matter for crunching, only for pushing many pixels). In the game benchmarks I've seen there was a surprisingly small difference between both versions. On the other hand we do know that GPU-Grid requires some memory bandwidth. Without actually testing it's impossible to say which difference it's going to make, though.

MrS
____________
Scanning for our furry friends since Jan 2002

The core is so different, that it is likely that CUDA3.0 does not work the GTX460.
Tomorrow I will updload the working cuda31 application.

gdf

Might be an idea to close Boinc and then open it again (wait for 10sec before opening it).
I did this before running a beta on my GTX470, and it finished OK; different card I know but might be worth a try!

PS might be better to post back to here.

The 460 came today.

I'll give it a go late tonight (after midnight) on both XP and Linux.

Although it does not work here yet, it still sounds like a very good card. The GF100's did not work straight out of the box either, so don't fret it.

Four times faster than a GT240, and using 160W! I have 4 GT240s in one system and they use 260W. The GT240 was as efficient as the bigger G200 cards, so these still look like a good buy.

What does Boinc report its performance as?

Yeah, the GF100 Fermi's are only CC2.0. I'm guessing the GF104 cards will better exploit CUDA 3.1 and 3.2.

That 363 GFlops is a joke though (a correction factor is called for there). It will do about 4 times what a GT240 does; they only have 96 shaders, whereas a GTX460 has 336 shaders and improved architecture!

- It should be 907, http://en.wikipedia.org/wiki/GeForce_400_Series

- 16/07/2010 23:50:01 NVIDIA GPU 0: GeForce GT 240 (driver version 19621, CUDA version 3000, compute capability 1.2, 512MB, 307 GFLOPS peak) - OC'd a bit but nothing special.

A system draw of 227 Watts running Collatz and 4 CPU projects at 100% is good going!
My Phenom II 940 with the 4 GT240s pulls 355W at the wall. You are doing about the same work for 36% less power.
Even my i7-920 (with Turbo off) uses 300 to 320W with my GTX470 running GPUGrid.

This is one of the failed GTX460 WU's:
Name 347-KASHIF_HIVPR_auto_spawn_2_90_ba1-26-100-RND1903_1
Workunit 1704860
Created 16 Jul 2010 20:03:26 UTC
Sent 16 Jul 2010 20:06:38 UTC
Received 16 Jul 2010 20:13:36 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status -40 (0xffffffffffffffd8)
Computer ID 67635
Report deadline 21 Jul 2010 20:06:38 UTC
Run time 1.450003
CPU time 0.920406
stderr out

<core_client_version>6.10.57</core_client_version>
<![CDATA[
<message>
- exit code -40 (0xffffffd8)
</message>
<stderr_txt>
# Using device 0
# There is 1 device supporting CUDA
# Device 0: "GeForce GTX 460"
# Clock rate: 0.81 GHz
# Total amount of global memory: 774307840 bytes
# Number of multiprocessors: 7
# Number of cores: 56
SWAN : Module load result [.fastfill.cu.] [200]
SWAN: FATAL : Module load failed

The core ratio to Multiprocessor is still out; it's 6 to 1 and should be 48 to 1:
7 Multiprocessors and 7 groups of 48 shaders (3x16).

Anyone care to drop the BOINC alpha mailing list a note that their number of multiprocessors are correct, but they have to multiply them by 32 for GF100 and by 48 for GF104 to get the correct number of shaders?

MrS

We will try to give a new application for the 460 on Monday.

gdf

Awesome! Can't wait to try it out.

Not having any luck with the 460 on Folding. It is running 3dmark 06 right now so the drivers must be working, correct?

@MarkJ: sorry, this incorrect reporting is an issue of GPU-Grid, not BOINC! SK reported this to them yesterday or so.. don't be surprised if they're a little p***** off. Just tell them it was my fault ;)

@GDF: will the correct number of shaders be reported in the new monday app?

@GTX460 & Collatz: don't forget that CC is a comparably light workload, i.e. the cards draw considerably less power here than under GPU-Grid (or Milkyway for ATIs).

Anyway, 14:30 is a nice result! THe 1 GB version may even improve this a bit, as CC loves memory bandwidth. For comparison: my HD4870 takes about 13 mins at 800 / 950 MHz core / mem. That's for 1.28 TFlops and 122 GB/s bandwidth. The GTX460 786 MB weights in at 0.9 TFlops and 86 GB/s. Apparently we're seeing slightly better utilization of the nVidia shaders here.

MrS
____________
Scanning for our furry friends since Jan 2002

Then Fermi cores and shaders are linked for both GF100 and GF104 card alike; we have to overclock both at the same time.

53C - I wish!

I'm running a GPUGrid WU that is only using 60% GPU (going by EVGA Precision) and my GTX470 is at 67'C. It would default to about 91deg, if I did not up the fan speed.

By the way anyone with a GF100 Fermi should increase the fan speed while crunching. There are 2 reasons:
1. The card stays cooler and should last longer.
2. When it is cooler it uses less energy (10 to 20W) - GF100's are leaky!
The power usage of running a fan faster is more than offset by the savings from leakage by the GPU. It will still leak, but not as much.
Think of it like an off shore oil well cap. Dont use one and it leaks everywhere. Only use one a bit and it still leaks, badly. Use a good one the correct way, and you have stemmed the flow as much as you can.

Keep the memory at stock. If you overclock the memory on the GTX460 it will end up slowing the card down due to increased errors (the controller is almost at its max as is, which is why they used 4000MHZ max. rated RAM rather than 5000MHz)!

Your Antec EarthWatts 380 is a good PSU, enough AMPS. I have one supporting a Q6600 and an OC'd GTX260-216, which uses more power than your GTX460.

I ran a Folding@home task again today, while running GPUGrid (on my GTX470), and the temps did not go up from 72degC (GPU Fan at 83% mind you). I think most apps are reporting the actual usage of GPUGrid incorrectly (or at least a skewed version). EVGA Precision reported GPUGrid as only using 60% GPU, but the temp was just the same, 72degC. When I suspend tasks it drops below 40degC. Running Folding it also (as with your 460 running Collatz) said 99% GPU usage, but the temp was just the same, 72degC.

By the way anyone with a GF100 Fermi should increase the fan speed while crunching. There are 2 reasons:
1. The card stays cooler and should last longer.
2. When it is cooler it uses less energy (10 to 20W) - GF100's are leaky!
The power usage of running a fan faster is more than offset by the savings from leakage by the GPU. It will still leak, but not as much.

Think of it like an off shore oil well cap. Dont use one and it leaks everywhere. Only use one a bit and it still leaks, badly. Use a good one the correct way, and you have stemmed the flow as much as you can.

Now it runs on gtx460 with acemdbeta 6.36.

gdf

Now it runs on gtx460 with acemdbeta 6.36.

gdf

You are now picking them up. Just let them run. You have driver 25896, which is good. For some reason you aborted one 6.36 Beta? Anyway hope the next one works for you.
PS. It's CUDA 3.1 (3010), not to be confused with CC2.1.

The full tasks are now available for your GTX460. They are called, ACEMD2: GPU molecular dynamics v6.11 (cuda31).
The Betas were called ACEMD beta version v6.36 (cuda31)

The full tasks are likely to take several hours, perhaps about 4h (just a rough guess) on a GTX460.

Is there any way to keep from downloading 3.0 tasks (which won't run) and only download 3.1 tasks?

Is there any way to keep from downloading 3.0 tasks (which won't run) and only download 3.1 tasks?

Is there any way to keep from downloading 3.0 tasks (which won't run) and only download 3.1 tasks?

trn-xs now has driver 25896 on Win7 x64 and is using a 6 core i7.
Yesterday morning he was downloading 6.05 Work Units. They all failed (obviously). Since the evening he only downloaded 6.11 and Beta tasks. He now has 3 tasks in progress, so I expect he removed one of his GTX460's, which would remove the possibility of power draw issues (not enough Amps). All Betas ran well, but the only 6.11 that was returned failed,

2696618 1716552 20 Jul 2010 9:03:03 UTC 21 Jul 2010 3:25:49 UTC Error while computing 9,380.86 8,030.28 4,428.01 --- ACEMD2: GPU molecular dynamics v6.11 (cuda31)

trn-xs, there seems to be lots of task restarts, so perhaps it is worth checking your Boinc settings - Perhaps you selected to not use the card when the user is active? Such start/stop/start/stop running slows the tasks down, a lot, and increases the chance of failures.

It's possible that one of the cards is error prone, or is just running too hot, but it is more likely that the buildup of heat in the case is causing overheating on other components. Did you check the card, case and CPU temperatures?
When one of these cards is in a system, it runs nice and cool, but when you pop two in, they reportedly start to get very warm (over 90degC), unless you manually turn the fan speeds up. The card nearest to the Northbridge tends to be hotter.
If your system was stable before you installed the cards and is stable now with one card, then the instability was probably down to using two cards; increased stress on system components causing crashes.

Still, it was a good idea to separate them. If you run one card for a day or two at stock, and then the other there are two likely outcomes:
- One card causes errors, in which case return it.
- Both run fine, in which case your hardware configuration was not up to the task.

If both work separately, its best to start by setting default clocks and improving the cooling; keep an eye on the temperatures and turn up the fan speed on the cards. You might well need another extractor fan, as these cards dissipate heat back into the case.

If that does not stabilize your system, look at other parts; check the PSU, RAM, for read/write issues.

Even though 220V is at the low end, that is a high end PSU designed to run GPUs. So 70A should be enough; at stock the cards will only draw about 27Amps between them running flat out(324W).

Is there any way to keep from downloading 3.0 tasks (which won't run) and only download 3.1 tasks?

If you a fermi and a 3.1 driver you should be downloading only 3.1 tasks. Isn't it the case?

gdf

That graph style is Anandtech, but I can't find any GP-GPU stuff in the 2 launch day reviews. Where is it from? I thought it was unsatisfying to bench games only, as there's not much to see there anyway..

MrS

Switched cards around so now the GTX 460 is running in XP64. GPU utilization went up from 49% to 95%. Times look to be much faster than before but still slower than my GTX 260. One would think that a more advanced 336 shader card should be faster than an old 216 shader card. The GTX 460 is also looking to be less than twice as fast as the 96 shader GT 240. Think there's still much work to do on the new app.

As a comparison with another client, Collatz is almost 3 times faster on the GTX 460 than on the GTX 260 and almost 5 times faster than on the GT 240. Of interest in terms of power draw, in the same machine the total draw is 175 watts with the GT 240 and 239 watts with the GTX 460, both running Collatz at 99% GPU.

What a leap in GPU utilization! Do you think the GTX470 might get the same kick in performance as the GTX460 if running XP 64bit? Skgiven said he had an XP 64bit somewhere & he has a GTX470 but didn't think that it was worth giving a try if he had to get more RAM & possibly a new license. How much RAM do you have & what CPU was used?

Bigtuna, your GTX 460 with 768MB using the 25896 driver and XP x86, is doing quite well for some tasks:
This TONI_CAPBIND for example took 15906 sec (6,803.41). Faster than this task, 17150 sec.
The more recent task is only 5% slower than a similar task, on my GTX260-sp216, 15095 sec.

My GTX260-sp216 is slightly factory overclocked and I have my shaders at 1525MHz, about the same as yours. My card is also on XP, and is supported by a Q6600 CPU and 4GB RAM. I have 1 CPU free and I am using swan_sync. So it is reasonably well optimised for GPU crunching.

Are you also using the swan_sync environmental variable, and do you have a core free?

I noticed earlier that there was a 10% difference between the 1GB and 768MB version you have. So what I am saying is that on XP a reference 1GB version might be slightly faster than a reference GTX 260 crunching TONI_CAPBIND tasks (about 5%). Dont know about the other tasks yet. Also your overclock from 1350MHz is not quite as much of a leap as mine, from 1242.

It would be nice to see a 1GB GTX460 on XP at 1600MHz, to see what they can really do.

Hi Bigtuna,

I've checked your PC & compared to the one Beyond uses. He has 2GB VS your 1GB RAM, but you've had lots of failed WU's, too few to make estimates (My personal opinion), Beyond has had lots of successful WU's.

But that's beside the point. What I wanted to know, was if the GTX470 which gpugrid has had more time to play with, also had a nice boost running XP 64bit compared to running XP 32bit.

The only way I'd guess how to make a comparison, was if someone who ran XP 32bit with a GTX470 installed XP 64bit w/o modifying his PC & shared that knowledge.

Are you also using the swan_sync environmental variable, and do you have a core free?

Did we ever decide exactly what swan_sync was?

So far I'm impressed and disappointed at the same time. The 460 runs cool and quiet which is good but the performance is only about 2.5 x GT-240 but the price is 4 x GT-240.

Well mine arrived today. I have swapped out a GTX275 for it and its off and running.

Interestingly GPU-Z 0.4.4 says it only has 224 shaders. Maybe they know something we don't :-)
____________
BOINC blog

Well initial performance is actually worse than the GTX275. As MJH said they need to work on the app. I'd say GPU-Z needs an update too.

Thats the price for rushing out and buying the latest & greatest toy. We just need to wait a while for the s/w to catch up.
____________
BOINC blog

Kill-a-Watt power info for the GTX-460. I've got the 768 MB version with a factory OC to 763/1526 MHz.

At idle the AMD 4000+ system draws 93 watts.

Running GPUGRID with Swan_Sync=0 the system draws between 195 and 210 watts.

I think it would be generically more accurate to compare the GTX460 768MB with the GTX260-192, and compare the GTX460 1GB with the GTX260-216:
The 192 version of the GTX260 was about 10% slower than the 216 version (at reference, for most applications and games that it worked on; not here).
The 768MB version of the GTX460 is about 10% slower than the 1GB version.

If the GTX460 cards are only using 224 of their 336 shaders then they square up fairly evenly in terms of shader performance to the GTX260, with the 1GB version perhaps being about 5% better than the GTX260-216 at reference (on XP or Linux). Obviously if you get a good overclocked 768MB version it could outperform a GTX280, but equally a good GTX260-216 with a high clock would outperform a native 1GB version of the GTX460.

The present GTX460 benefits for crunching here are that they are quieter, run cooler, use less electric, and overclock more.
The problems are that they are still slightly too expensive, there are not enough 1GB card available, the drivers just really got the card working and need to mature, as does the app (after the holidays).

Come the autumn (and the GTX475 with 8cores and 384shaders) there will likely be new drivers and a new app, which should improve performances. The CC1.3 cards are almost 2 years old, so the drivers and apps are well refined at this stage; the present 6.05 app is well over twice as fast as the original.

It is likely that the power usage will rise slightly once these cards are better utilized, and the performance will increase somewhat more, so it’s a bit too early to do an accurate crunching performance per watt comparison here. The good news is that the architecture is set to remain for the GTX475, so when that turns up, the scientists will be better prepared for it.

The shader speed is more important here than the RAM speed. Besides, the actual speed of the GDDR5 is not the issue with the 768MB card versions, the issue is the memory interface!

"The more expensive GTX 460 1GB version is based on the full-fat GF104 GPU and so has the maximum 256-bit memory interface. This is because the 1GB version of the GTX 460 has more memory links, and therefore more memory bandwidth. While the GTX 460 768MB has only six links to memory, so has only six 128MB memory chips on the card (6 x 128MB = 768MB), the GTX 460 1GB GPU has the full eight connections (8 x 128MB = 1GB). As each memory connection is 32 bits wide, the GTX 460 768MB has a 192-bit memory interface (6 x 32 bits = 192 bits) while the GTX 460 1GB has a 256-bit bus". Bit-tech, by James Baker.
Many games also observe this 10% reduction in performance between the 1GB and 768MB versions.

I see the GTX460 as performing about the same as a GTX260, with the 768MB version falling slightly behind the GTX260 and the 1GB version slightly outperforming a reference GTX260. There are plenty of results to go by.

@bigtuna: What was the percentage of GPU usage for your Kill-A-Watt readings

Did we ever decide exactly what swan_sync was?

It controls the method that ACEMD tells the CUDA runtime to use for polling for GPU work completion. The default method is for the application to block until completion; this keeps CPU load to a minimum but introduces latency that slows the program down.

Setting SWAN_SYNC=0 will cause ACEMD to poll for kernel completion, which minimises latency at the cost of CPU.

0 is the only valid value for SWAN_SYNC - anything else will cause undefined behaviour, so don't do it!

MJH

I don't think the 1Gb cards are going to provide any better bang for the buck.

According to GPU-Z the memory controller load when running GPUgrid 6.11 app and Swan_Sync=0 is around 15 to 17% on my 768Mb card. Its hardly under pressure.
____________
BOINC blog

2.8k/day is more points that that one core would pull down running Rosetta.

But it's less than 1 core can get in Aqua@home hence why i keep my gtx 260 without swan_sync.

Same here, and CPU credits aren't exactly comparable to GPU credits. Some projects can only be run on the CPU and a core on those projects means a lot.

On the other hand lets compare credits between the GTX 260 and GTX 460. My GTX 260 does around 40,000 credits/day in GPUGRID and about the same or even a bit less in Collatz. On the other hand my GTX 460 does around 35,000 credits/day in GPUGRID and about 103,000 credits/day in Collatz without sacrificing a CPU core. Quite a difference...

If it works, I would expect that all of them will benefit even g200 cards, but GF100 more and GF104 even more.

gdf

IMO the only way as not to discriminate or favoritise projects, is if BOINC Projects can agree to give credits based on CPU/GPU performance per watt & the CPU/GPU time used on WU's x deadline bonus. Is this even possible though???

If effective CPU/GPU's were encouraged & ineffective CPU/GPU's were discouraged. BOINC could send the message that wasting watts is a waste of time.

CTAPbIi, at present your overclocked GTX275 is supported by mature drivers and 2 years of application development. It outperforms most GTX460's by about 25%. The GTX460 has only been released and will take time to optimise, after the summer holidays.

I don’t see the GTX460 as an upgrade path for someone with a GTX275. It would be a step up to a new generation of card but a step down in terms of relative performance. Wait and see how the GTX475 turns out - it's due out in the autumn. It's not as if you are desperate to replace a failing, highly inefficient or obsolete card.

Even if the GTX460 prices fell or the GTX475 turned up early there would be little point replacing a faster card with a slower one. Enough people have a GTX460 for the scientists to try out new applications. After the GTX460 is better supported, it may or may not be faster than your card. It could take several application updates and driver releases before the GTX460 is a match. Your GTX275 is a good card and will be for some time yet. The shops can wait.

My general advice is: don't rush to buy anything, sit back and remember it's only a hobby after all. But if you see the opportune moment - go for it (e.g. some relative or friend needs a new card anyway). And if you buy go for best bang for the buck - but also remember that you only have so many PCIe slots and that running a PC to feed the GPU also costs power. I'd rather have 1 24/7 PC crunching with a fast GPU than 2 PCs with 2 slower GPUs.

MrS
____________
Scanning for our furry friends since Jan 2002

i jusr read the topic quickly. correct me if i'm wrong, but if i've got OC'd GTX275 there is no sense to buy GTX460?

"With Swan_Sync "on" (Syan_Sync=0) similar tasks are taking about 15,850 seconds"
right now it takes me 14,550 (http://www.gpugrid.net/workunit.php?wuid=1737805)

the only thing - i'm using linux.

So, guys, what's the answer - should I rush to the shop or not? or may be it's better to wait a bit for GTX475 with 384 cores? what's the latest rumours - when we can expect it? later the summer?