Mine seems pretty responsive, maybe a touch slower than the old app. Overall, I am pleased so far.
PS I don't have the latest/greatest hardware out.

It is slightly slower. Most of the slowdown is since the likelihood at the end isn't currently done on the GPU which I'm planning on fixing.

But even if you do the likelihood on the GPU, it won't be nearly enough to comepensate the total performance loss of 25% compared to the old app.

Adding it all up here across the machines, i roughly lose ~100k credits/day using the new app.

---

Join Support science! Joinc Team BOINC United now!

[But even if you do the likelihood on the GPU, it won't be nearly enough to comepensate the total performance loss of 25% compared to the old app.

Adding it all up here across the machines, i roughly lose ~100k credits/day using the new app.

I'm not seeing this drop. I was seeing it in the 0.58/0.59 if you had a high CPU load, but this release fixes nearly all of that.

[But even if you do the likelihood on the GPU, it won't be nearly enough to comepensate the total performance loss of 25% compared to the old app.

Adding it all up here across the machines, i roughly lose ~100k credits/day using the new app.

I'm not seeing this drop. I was seeing it in the 0.58/0.59 if you had a high CPU load, but this release fixes nearly all of that.

That's on my radeon 5870. Things might be different on other GPUs.

Mine seems pretty responsive, maybe a touch slower than the old app. Overall, I am pleased so far.
PS I don't have the latest/greatest hardware out.

It is slightly slower. Most of the slowdown is since the likelihood at the end isn't currently done on the GPU which I'm planning on fixing.

But even if you do the likelihood on the GPU, it won't be nearly enough to comepensate the total performance loss of 25% compared to the old app.

Adding it all up here across the machines, i roughly lose ~100k credits/day using the new app.

I'm seeign about a ~7 seconds increase compared to 0.23 on my 5970. Time has increased from ~83 to ~90 seconds for a st'd wu, but for a longer wu the increase in time is closer to 9 seconds. So I'm not seeing a 25% slow down.

[But even if you do the likelihood on the GPU, it won't be nearly enough to comepensate the total performance loss of 25% compared to the old app.

Adding it all up here across the machines, i roughly lose ~100k credits/day using the new app.

I'm not seeing this drop. I was seeing it in the 0.58/0.59 if you had a high CPU load, but this release fixes nearly all of that.

Sorry to spoil the fun but I still see high system use (25% on a quad) while the system response is a little better.

I started playing with the polling mode without luck:
<=40 give high gpu load, but comes with the system use mentioned above
> 40 give gpu use of 75& and less, but keeps the system use nearly invisible
It's either high system use or low gpu use.

Is there some relationship between polling mode and target frequency that I need to take care of?

System is win xp 32bit, catalyst 10.10, HD5850

It smells like a problem in the communication between cpu and gpu. Either the cpu hangs in a wait and show it as system cpu time (not mw app cpu time) slowing the system unreasonable down or the gpu isn't getting enough work.

Reading most people have no problems with cat 11.3 I will try that one and see if this changes anything.

Not really a change with the new catalyst version. 11.3 feels a few seconds slower than 10.10.

Different size WUs seems to have different sweet spots.
With some of them (type 10_3 I think) I saw
polling mode 32: gpu 98%, cpu (system) 25%,
polling mode 33: gpu 80%, cpu (system) 3%,
The mw app still shows only a few seconds cpu time used.

[But even if you do the likelihood on the GPU, it won't be nearly enough to comepensate the total performance loss of 25% compared to the old app.
Adding it all up here across the machines, i roughly lose ~100k credits/day using the new app.

I'm not seeing this drop. I was seeing it in the 0.58/0.59 if you had a high CPU load, but this release fixes nearly all of that.

That's on my radeon 5870. Things might be different on other GPUs.

Comparing the performance on my 4770, 5850 and 5870 cards: I'm seeing less than a 5% slowdown in .62 from .23 on all 3 GPU types. That's with both apps tweaked to run at 99% GPU.

Great improvement Matt - Thank You for the efforts - not easy developing under these conditions/pressures.

Running 2x5970s on a 1090T @ 3.7Ghz 16Gb RAM with Version 0.62 run at default settings, no tweeking, running @810/500, Catalyst/Drivers 11.3.

CPU use is about 20% up on 0.23 - real world use right now is fine, not a problem, I did have big CPU use before.

GPU Utilisation is much better, peaks at 94-97%. It vary hugely and can go as low as 45%, does not stay at peak for too long. Temperatures therefore still very cool for a MW application. On the surface at this end looks like still have the capacity to increase utilisation in parts of the application. It will spend most of its time in the high 80's low 90's within the bracket 76-97%

No screen lag, responsive on mouse and keyboard. Can open browsers etc on top of it fine.

Speed is around 5% slower than 0.23, which would tie in with the slightly reduced utilisation level on 0.62 compared to 0.23

I had four validation/invalid errors initially, all my fault, since sorted. Otherwise, all seems to be going well. Am now confident enough to let it run for an extended period until lunch tomorrow and see what happens, on current showing I'm sure it will be fine.

Thanks again, your efforts are paying off, it continues to improve, and 0.62 is a Production Runner right now on initial impressions, we then take on board any improvements that may come along in time.

Dont forget to sleep .... :)

Regards from a grateful cruncher
Zy

My site is updated once again.

---

Good Job - Thank you Matt.

My 2 HD4850 together with its lonly AMD Athlon(tm) 64 Processor 3500+ single core are working fine again. (I even get 12 WU's at a time, thats 100% more).
Performance is now about 5 % lower then with the 0.23 but on the other Side, the GPU Temperature is not so high, and the system with its old singel CPU is more responsive.
This all with the stock app and no special options.
On WE I will play with the opti apps.

thx

franz

0.62 now been running 8hrs unattended, still fine, no errors or invalids, crunch times are ok, latter no doubt will steadily increase slightly as its tweeked overtime. My second machine running 1x5850 has joined the party now, same result as above for the 5970s on my primary machine.

Looking like a production runner - judging by the response on 5970s & 5850s at the least - then follow with incrementals over a more relaxed timeframe now it seems to be pretty well there with the Major parts of it.

You probably got a list as long as your arm of things you want to do with it, always the way I guess, but really at this end, its fine in the guise of an initial production runner. If you do manage to keep it running this cool still, as you finalise the tweeks and small bits, that would be real magic, a major plus. Its running nice and cool, and if it stays that way after final optimisation/utilisation, you'll gain a large chunk of new crunchers. Its a definite thumbs up :)

Regards
Zy

My site is updated once again.

Edit

@Arkayn - downloaded and running. I'm starting to see maybe 15% improvement over stock. Is this about right?

Two thumbs up!!

0.62 ran good:

Stock app, no app_info.xml.

Majority WUs ran at 98% GPU from start to finish.
0% CPU use, 99% CPU by AQUA-fp-WU.

Certain WUs dipped to 80%-90% GPU at T-50% or T-25secs for ~3secs then went up to 98% GPU.

At start: negligible ~2secs 25%-40% CPU.

At end: ~5secs 50% CPU.

Run time is almost similar to 0.23.

ATI-5850 (11.2), C2D E6600, 8GB RAM, Win7x64-ultimate.

Good work!

No screen lag, responsive on mouse and keyboard. Can open browsers etc on top of it fine.

Unfortunately, my experience under WinXP with Catalyst 11.3 is just the opposite. 0.62 slows the screen down so much it renders the machine almost unusable. What changed since 0.59?

Brian, on two of your machines all your tasks are showing as 0.50 SSE2 when we are now on 0.62?

My two (underclocked) 4850s scored 107K yesterday, down a little bit over 20% compared to before the current mess started. My client has been automatically downloading new apps; last I looked (yesterday) it was v.62. Last I looked (yesterday,) v.62 on my two 4850s still needed more than one core (on my two processor dual core Operton system) to reach 99% GPU utilization. At least now MW is yielding to my preferred non-MW CPU work units.

---

Please: WCG - Help Cure Muscular Dystrophy

No screen lag, responsive on mouse and keyboard. Can open browsers etc on top of it fine.

Unfortunately, my experience under WinXP with Catalyst 11.3 is just the opposite. 0.62 slows the screen down so much it renders the machine almost unusable. What changed since 0.59?

Abort all the 0.50's then do a detatch/reattatch, that should ensure you have an uptodate bunch of files - see how it goes then.

EDIT: oppps ..... 0.50's are the cpu ones ..... ignore comment .... sorry, bad hair day :)

Regards
Zy

No screen lag, responsive on mouse and keyboard. Can open browsers etc on top of it fine.

Unfortunately, my experience under WinXP with Catalyst 11.3 is just the opposite. 0.62 slows the screen down so much it renders the machine almost unusable. What changed since 0.59?

The polling mode. You can configure it with the options I mentioned. Which GPU is this?

The apps I have on my site are just stock, just with the included app_info.xml file and Matt's configure options.

---

Brian, on two of your machines all your tasks are showing as 0.50 SSE2 when we are now on 0.62?

The third machine (ADMIN2) is the XP machine with the HD5870. The other two have only HD5770's that can't be used for MW.