So here is how it seesm to me. I'm a car guy so I tend to use car analogies...
{
Was wrenching on my Mustang today and hoped in the r32 for a parts run. As I shifted her from 2 to 3 at 12,000rpm a thought occurred to me..

The mustang would blow up over say 7k but she makes 400ft lbs at 2000. Both the Ford and Volkswagon are about 440 horse but they are very different animals. I can't say one is better as I love them both. Each excels in their own way.

Anyway, so this is my analogy, long winded and confusing as it may be:

The Nvidia card has a lot of torque and low revs, The ATI revs high but a short little stroke.

Remember: Horsepower keeps you going..but torque,
.it makes you go..

I prefer the Torque. :D You got to be able to hook that power to the ground. Otherwise you just sit and spin.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Then you would like the power and torque of a 5.2 litre V10 VW Tourag diesel - 513 bhp, and 850 lbs foot of torque.

O - 62 mph in 7.9 seconds, and top speed limited to 140mph.

Mind you they do make quicker and more powerful diesels as well.

Then you would like the power and torque of a 5.2 litre V10 VW Tourag diesel - 513 bhp, and 850 lbs foot of torque.

O - 62 mph in 7.9 seconds, and top speed limited to 140mph.

Mind you they do make quicker and more powerful diesels as well.

Not bad for a diesel. Big turbo on it? My cars got a 455 w/a B&M th400, estimated ~375+ horse & 450+ torque. And it's still fairly mild. :P Hooks up good on the street. Fun to drive. :D

Wouldn't mind building a 400 with a 455 crank. Easy to get a lot of torque.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

I have a few Nvidia cards over on Seti..A 260 and a Tesla c1060.

What is the difference in speed?

I need to rerun my test, but, when I ran a bunch of work with the then application on my 9800GT it took the same time as the GTX280 card I had... not sure if it was me, or my stop watch, or what ...

I know that the 280 and my 295 are about 2-3 times faster than the 9800 on GPU Grid and I was pondering if the C1060 would be worth the effort ...

Last question, is it single core? Or double like the GTX 295?

OK, here's my scores on the doors for ATI overclocking;

HD4870; GPU clock: 800MHz, Memory Clock: 900MHz, Temperature: ~80C

HD4850; GPU clock: 685MHz, Memory Clock: 750MHz, Temperature: ~85C

HD4850; GPU clock: 730MHz, Memory Clock: 750MHz, Temperature: ~86C (Zalman fan)

In all cases I'm keeping the Memory Clock down to the lowest setting to keep the heat down, conserve power, and because I don't need it.

The first two have been stable for some days. The third only just came on line and needs some more testing, although I have tested it previously in another PC.

---

Then you would like the power and torque of a 5.2 litre V10 VW Tourag diesel - 513 bhp, and 850 lbs foot of torque.

Useless for MW though until they do a Double-Precision gearbox...

Howzit Paul?
Umm never really timed them, and the angle range thingy, with teamwork the task might crunch on all or more than one gpu etc. All the Nvidia cards stay in the same machine currently.

Also, here is the thing, my card was an engineering sample and therefore Nvidia only put three clusters in it so unlike the production unit mine only has 128 procs and 1g.

Near as I can tell it does units at least twice as fast as my 192core 260. The 260 takes about 1.3 steps per percent to complete in boinc manager, love my precise syentific measurments *chuckle*, The c1060 does more than 1% per step. The cpu use in task manager goes between maybe 0-5% for the 260 being 1% most often. The 1060 uses up to 8 percent and rarely drops below 5. Also where as the 260 runs about 124F the Tesla never gets over 112-114 and is rock solid it doesn't fluctuate ie. cool down between tasks like the others. Oh, the only other thing is my Tesla is built on a 200 board and I think the production one is an 8800 or some such.

Anyway, if can give direction for finding task time besides stop watch (I bored fast) would be happy to share. Quite curious. Waiting for rac to level off only just finally got in some good days as I had a buggy PS that caused me endless grief the first few weeks but she's working on 4 days uptime now so all is well. Cuda rac for that machine just passed 4,400 :D

Shoots huh, Aloha and keep crunchin'!


edit:

Then you would like the power and torque of a 5.2 litre V10 VW Tourag diesel - 513 bhp, and 850 lbs foot of torque.

How much is that in FLOPS?

Useless for MW though until they do a Double-Precision gearbox...

Is that like a two-speed diff? ;)

---

Howzit Paul?
Umm never really timed them, and the angle range thingy, with teamwork the task might crunch on all or more than one gpu etc. All the Nvidia cards stay in the same machine currently.

I timed them using the wall clock that I face. WIth the GTX280 and 9800GT I was seeing a 9 minute time per SaH task. I think that they have now added more output to the task file so that you can get more correct timings there.

What I suspect is that SaH has only a need for "x" parallel units and adding more does not do anything for the speed. Other tests have indicated that there is a timing difference between the various cards. But, as in everything, YMMV ...

My RAC for SaH is going below 1 soon and I will have to do some more tasks if I want to post there later on ... though ... I don't think ... that I will be in any big hurry to do so ...

But the reason for my question is that I will be migrating towards only 3 windows system and with that I need to maximize my computing power. I am idly considering moving to systems with 3 PCI-e slots with a configuration of something like a GTX295, C1060, and an ATI GPU HD4890(2) ... Of course, I have yet to hear of anyone running both an ATI and Nvidia card in the same system. Since I don't have spare cards and am not setting up a new system ... well, that is why it is idle speculation ... :)

I have actually switched off all rigs except the Nvida and Ati crunchers. Much easier on the electric with only 2 rigs. I don't mean the electricity they use directly but the major cost was in running the AC to try to keep it only near 80f in here. Envy those folks who can put puters in their cold garage. But can't take the cold myself so not moving anytime soon.

---

I have actually switched off all rigs except the Nvida and Ati crunchers. Much easier on the electric with only 2 rigs. I don't mean the electricity they use directly but the major cost was in running the AC to try to keep it only near 80f in here. Envy those folks who can put puters in their cold garage. But can't take the cold myself so not moving anytime soon.

That was my original plan. One laptop with nvidia for using, one ATI for crunching; turn off all other including dilapidated old PCs - save on electricity and micromanaging a number of PCs. But then I had to have another ATI... and another...

---

But then I had to have another ATI... and another...

Your an addict. You need to goto MilkyWay Anonymous. :P

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Cathryn's Clinic will help, and nurse Olga will be the one to give you the emetic before a forceful colonic irrigation.

130,000+ RAC with 3 ATI cards. How many PCs are you using to get a fraction of this?

It's not me that needs help...

---

130,000+ RAC with 3 ATI cards. How many PCs are you using to get a fraction of this?

It's not me that needs help...

Pull down a few uFluids workunits and see if that affects your ATI crunching. The darn thing killed my CUDA processing for some reason.

---

me@rescam.org

I fantasize that this will help me overclock all my NVidia GPUs for a faster and hotter cup of green china -

Your an addict. You need to goto MilkyWay Anonymous. :P

Actually, I thought it was the addicts who run their machines hidden as anonymous.. ;)

---

130,000+ RAC with 3 ATI cards. How many PCs are you using to get a fraction of this?

It's not me that needs help...

Pull down a few uFluids workunits and see if that affects your ATI crunching. The darn thing killed my CUDA processing for some reason.

I may just take those fluids on. But right now I need some toast and a visit see how my pond fish are doing in their fluids.

---

I'm eagerly waiting for the release of the HD4890 in just a few weeks. The faster frequencies should make for great performance here.

130,000+ RAC with 3 ATI cards.

I managed to stay ahead of your RAC with two ATI cards and a fraction of a few CPUs for some time - not so after a PSU died and took the mains breaker with it yesterday while I was away. Back in business today after an emergency wallet blood-letting, but I suspect 3 cards can be tough to beat anyway.

But your point is obvious, and one that I'm told some people who do professional graphics rendering also have understood: Let the graphics card do the heavy work and get more speed for less money.

---

130,000+ RAC with 3 ATI cards.

I managed to stay ahead of your RAC with two ATI cards and a fraction of a few CPUs for some time - not so after a PSU died and took the mains breaker with it yesterday while I was away. Back in business today after an emergency wallet blood-letting, but I suspect 3 cards can be tough to beat anyway.

But your point is obvious, and one that I'm told some people who do professional graphics rendering also have understood: Let the graphics card do the heavy work and get more speed for less money.

I spent a while trying to catch you when I had two ATI cards - the third I added very recently, just yesterday ;)

---