Big Unit moves in!

[BakedON]

3.2ghz actually.....
And it seems to have gravitated down to 17m / frame... and will probably settle even lower than that. The first few frames got my hopes up.

DUH! When I'm USING the machine the client slows down!
when I walked away for a half hour speed went back up to 11m27s / frame

 

[BillR]

DUH! When I'm USING the machine the client slows down!
when I walked away for a half hour speed went back up to 11m27s / frame

Unless you have the ultimate magic Intel 3.2 I really don’t think you mean 11m 27s per frame. Since a 600 point WU has 100 frames that would give you the fastest computer Intel has ever built I believe. My math could be bad…but my Intel isn’t quite that fast and it only folds.

 

[Killer[MoB]]

Unless you have the ultimate magic Intel 3.2 I really don’t think you mean 11m 27s per frame. Since a 600 point WU has 100 frames that would give you the fastest computer Intel has ever built I believe. My math could be bad…but my Intel isn’t quite that fast and it only folds.

Right on Bill. LOL

Actually, it's probably just whatever he is using to get the frame times is not reporting them correctly. I'm sure he'll get that staight for us.

The fastest 2 CPUs on my list (of 30 samples) are a Opteron Hammer @ 2.4 and an A64 939 Winchester @ 2.5. Then a P4 @ 3.6 and A64 754 Newcastle @ 2.46 are 3rd and 4th.

Times are 27.91, 27.94, 28.5 and 28.71 minutes per frame respectivly.

 

[Hito Bahadur]

Right on Bill. LOL

Actually, it's probably just whatever he is using to get the frame times is not reporting them correctly. I'm sure he'll get that staight for us.

The fastest 2 CPUs on my list (of 30 samples) are a Opteron Hammer @ 2.4 and an A64 939 Winchester @ 2.5. Then a P4 @ 3.6 and A64 754 Newcastle @ 2.46 are 3rd and 4th.

Times are 27.91, 27.94, 28.5 and 28.71 minutes per frame respectivly.

I'd like to see what a high Mhz Pentium M would do.

 

[Killer[MoB]]

I'd like to see what a high Mhz Pentium M would do.

Me too. Your 1.8M is the highest PpHpGHz on the list. Very impressive.

 

[Mauli]

41 minutes per frame on the 2600+ @ 2.08 sounds right in line with where it should be.

50 minutes is way off for the A64 @ 2.25 though. You should be seeing times of 31:25 to 32:00 or so per frame when the computer is only folding.

yeah i was thinking that my a64 3400+ should perfom better, any idea what would cause that?

 

[BakedON]

Unless you have the ultimate magic Intel 3.2 I really don’t think you mean 11m 27s per frame. Since a 600 point WU has 100 frames that would give you the fastest computer Intel has ever built I believe. My math could be bad…but my Intel isn’t quite that fast and it only folds.

THAT'S where the problem is.... F@H lists it as 100 frames but the client here is seeing it as 250 frames. Before being all mean (I know... it's just how you are ya big bully ) you should have seen that while I posted 11m27s/frame I ALSO posted an estimated completion in 01:22:17:00... if you had done the math you'd see that 100 frames at 11.5m per is only 19-someodd hours.
All that being said... I hope there's nothing wrong with this WU.... I'd hate to get it done just to have it end up being worthless to F@H.....

 

[Tigerbiten]

yeah i was thinking that my a64 3400+ should perfom better, any idea what would cause that?

Check in the FAHlog file that your running with SSE boost.
No boost is one reason for it to be running aprox 100% slower than it should.
Are you useing the -forceasm switch.

Luck.........

 

[Mauli]

Check in the FAHlog file that your running with SSE boost.
No boost is one reason for it to be running aprox 100% slower than it should.
Are you useing the -forceasm switch.

Luck.........

It says SSE Boost ok.

And what is forceasm

 

[Bait-Fish]

Heh, despite my bellyachin' a couple days ago I've since had three of these load up on three of my boxen. Foldin' on!

 

[BillR]

THAT'S where the problem is.... F@H lists it as 100 frames but the client here is seeing it as 250 frames. Before being all mean (I know... it's just how you are ya big bully ) you should have seen that while I posted 11m27s/frame I ALSO posted an estimated completion in 01:22:17:00... if you had done the math you'd see that 100 frames at 11.5m per is only 19-someodd hours.
All that being said... I hope there's nothing wrong with this WU.... I'd hate to get it done just to have it end up being worthless to F@H.....

Just got my first biggun.
p1134_RIBO_FSpeptide_EXT
11m27s / frame
01:22:17:00 estimated to complete

Not bad....

The p1134 is a 100 frame 600 pointer. That's what got me to thinking.

Wasn't being a bully, just my usual sarcastic self

 

[Tigerbiten]

And what is forceasm

"C:\Tigger@Home\F@H1\FAH500-Console.exe" -forceasm -local
This is one of my clients shortcut properties

-forceasm is one of the switches that you can add to your shortcut.
If the client crashs or shuts down wrong then it sets a flag which turns off the boost.
The switch overrides the flag so the boost is always on.
It means I dont have to reboot all the clients after a power cut.

-local is because it from one of my duallies.
It forces it to read the local config file so it gets the machine ID right.

Other switches.

-config / -configonly. If you want to change any settings on the console version.

-advmethods. Will turn the client to a Beta testing platform. If You run a FAH client with this switch then occasionally/constantly this client will be fed with experimental Cores and WUs so at the worst case there may be (some?) crashes and degradation of received points.

Ps. Addanother one. Rabbit. p1135, 49:03, 3.49 PpHpG.

Luck............

 

[unhappy_mage]

What CPUs are in those boxen? Thanks in advance!

All 3 are athlon 900s with at least 384MB of ram. Sorry I forgot to mention that.

And my fast boxen have smaller gromacs... thanks stanford.

 

[DF454]

My main box is just finishing up the first 600 pointer I've seen.

P1134 6/19/0 156

Time/Frame: 00:38:10
Time/WU: 63:36:40

This is on a XP Barton 2500 @ 2050.

 

[Bodega]

heh... lost one on the 99th frame

 

[marty9876]

heh... lost one on the 99th frame

Not cool, I got over 20 of these buggers in the farm right now. I really hope there are stable, that's a lot of something if they don't get done.

Great, now I'm worried....



Marty

 

[Arkaine23]

"C:\Tigger@Home\F@H1\FAH500-Console.exe" -forceasm -local
This is one of my clients shortcut properties

-advmethods. Will turn the client to a Beta testing platform. If You run a FAH client with this switch then occasionally/constantly this client will be fed with experimental Cores and WUs so at the worst case there may be (some?) crashes and degradation of received points.

This is not quite correct.


-betateam will get you beta work units, and you will be expeceted to give feedback in the hidden folding community forums beta section about your beta wu's.

-advmethods gives you projects which are "freshly released from beta testing", it will also steer your client to grab big packet work if big units are available and your client.cfg is set for big packets. Big units supply a 100% boost in points over their benchmarked value, they use much more ram to process and are larger to upload/download.


Admittedly, in either case you sometimes get unstable projects that will fail to complete and can either be lost entirely or returned partially finished for partial credit.

 

[Mauli]

"C:\Tigger@Home\F@H1\FAH500-Console.exe" -forceasm -local
This is one of my clients shortcut properties

-forceasm is one of the switches that you can add to your shortcut.
If the client crashs or shuts down wrong then it sets a flag which turns off the boost.
The switch overrides the flag so the boost is always on.
It means I dont have to reboot all the clients after a power cut.

-local is because it from one of my duallies.
It forces it to read the local config file so it gets the machine ID right.

Other switches.

-config / -configonly. If you want to change any settings on the console version.

-advmethods. Will turn the client to a Beta testing platform. If You run a FAH client with this switch then occasionally/constantly this client will be fed with experimental Cores and WUs so at the worst case there may be (some?) crashes and degradation of received points.

Ps. Addanother one. Rabbit. p1135, 49:03, 3.49 PpHpG.

Luck............

might be a stupid question, but how does that make it faster? I am aiming to get my cpu to fold faster... for the record, my amd64 doesnt fold as quick as its supposed to

 

[Killer[MoB]]

yeah i was thinking that my a64 3400+ should perfom better, any idea what would cause that?

Is it possible something else is running that keeps folding from getting all of the CPU? When you are not doing anything else on the computer, folding should pull from 98 to 99% of your CPU cycles.

I would figure that you would have at least 512 megs of ram on that 64, so that should not be a problem.

What OS are you running?

Will be back on the forum after Lost.

 

[Arkaine23]

-advmethods = helps get the best PPD WU when used with big packets enabled in client.cfg and enough ram on an SSE capable CPU

-forceasm = keeps SSE/3dnow turned on for gromacs WU's even if you have an improper shutdown of the folding client

 

[Mauli]

Is it possible something else is running that keeps folding from getting all of the CPU? When you are not doing anything else on the computer, folding should pull from 98 to 99% of your CPU cycles.

I would figure that you would have at least 512 megs of ram on that 64, so that should not be a problem.

What OS are you running?

Will be back on the forum after Lost.

Winxp pro sp2 , a64 3400+, 1gb ocz plationum kv8-max 3 mainboard

 

[Hito Bahadur]

Winxp pro sp2 , a64 3400+, 1gb ocz plationum kv8-max 3 mainboard

Mauli,

Have you verified the actual frequency of your rig? Through boot-up, sandra, etc. It sounds like you might be underclocked for some reason.

 

[Mauli]

Mauli,

Have you verified the actual frequency of your rig? Through boot-up, sandra, etc. It sounds like you might be underclocked for some reason.

"My Computer" tells me so, so does the bios. But ill go right ahead and cehck with sandra

 

[Mauli]

This is what i got:

SiSoftware Sandra

Benchmark Results
Dhrystone ALU : 9921 MIPS
Whetstone FPU : 3576 MFLOPS
Whetstone iSSE2 : 4636 MFLOPS

Performance Test Status
Run ID : MAULI on Donnerstag, 6. Januar 2005 at 20:20:34
NUMA Support : No
SMP Test : No
Total Test Threads : 1
SMT Test : No
Dynamic MP/MT Load Balance : No
Processor Affinity : P0
Number of Runs : 64000 / 640

Processor
Model : AMD Athlon(tm) 64 Processor 3400+
Speed : 2.25GHz
Model Number : 3400 (estimated)
Performance Rating : PR3371 (estimated)
Type : Standard
L2 On-board Cache : 1MB ECC Synchronous, Write-Back, 16-way set, 64 byte line size

Chipset 1
Model : Abit Computer Corp Apollo K8HTB CPU to PCI Bridge
Front Side Bus Speed : 2x 816MHz (1632MHz data rate)

Chipset 2
Model : Advanced Micro Devices (AMD) Athlon 64 / Opteron HyperTransport Technology Configuration
Front Side Bus Speed : 2x 816MHz (1632MHz data rate)

Features
SSE2 Technology : Yes
SSE3 Technology : No
HTT - Hyper-Threading Technology : No

Performance Tips
Notice 5008 : To change benchmarks, click Options.
Notice 5004 : Synthetic benchmark. May not tally with 'real-life' performance.
Notice 5006 : Only compare the results with ones obtained using the same version!
Tip 233 : Consider using the Win64 x64 version of SiSoftware Sandra.
Tip 2 : Double-click tip or press Enter while a tip is selected for more information about the tip.

edit:

This is my amd 2660+

SiSoftware Sandra

Benchmark Results
Dhrystone ALU : 8481 MIPS
Whetstone FPU : 3226 MFLOPS

Performance Test Status
Run ID : FILEBANK on Donnerstag, 6. Januar 2005 at 21:00:39
NUMA Support : No
SMP Test : No
Total Test Threads : 1
SMT Test : No
Dynamic MP/MT Load Balance : No
Processor Affinity : P0
Number of Runs : 64000 / 640

Processor
Model : AMD Athlon(tm) XP 2600+
Speed : 2.08GHz
Model Number : 2600 (estimated)
Performance Rating : PR3016 (estimated)
Type : Standard
L2 On-board Cache : 256kB ECC Synchronous, Write-Back, 16-way set, 64 byte line size

Chipset 1
Model : ASUSTeK Computer Inc nForce2 AGP Controller
Front Side Bus Speed : 2x 166MHz (332MHz data rate)

Features
SSE2 Technology : No
SSE3 Technology : No
HTT - Hyper-Threading Technology : No

Performance Tips
Notice 5008 : To change benchmarks, click Options.
Notice 5004 : Synthetic benchmark. May not tally with 'real-life' performance.
Notice 5006 : Only compare the results with ones obtained using the same version!
Tip 2 : Double-click tip or press Enter while a tip is selected for more information about the tip.

 

[Hito Bahadur]

This is what i got:

Looks good. The only other thing I can think of is temperature throttling.

 

[Mauli]

Looks good. The only other thing I can think of is temperature throttling.

meaning that my temp is too high?

 

[Hito Bahadur]

meaning that my temp is too high?

Yes a.k.a Cool'n'Quiet/PowerNow!

There are some utilities here:

http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_871_9706,00.html

that can help trace whether throttling is going on and can also force the CPU to be running at full speed all the time.

 

[SmokeRngs]

My system finally put it's first 600 pointer in. Got a nice jump in points. Now it's working on another 300 pointer.

 

[Leadman584]

OK, what gives. I installed console version of F@H on my lappy, because it failed to acquire a new work packet, using graphic interface. Did not enable big packets for obvious reasons. Just checked it, and it's 98 steps into a 600 pointer. I did something wrong.

Man these things run slow on an AMD M 3000+ rig. 3.0+ P4's blast through these things a whole lot faster.

 

[Hito Bahadur]

OK, what gives. I installed console version of F@H on my lappy, because it failed to acquire a new work packet, using graphic interface. Did not enable big packets for obvious reasons. Just checked it, and it's 98 steps into a 600 pointer. I did something wrong.

Man these things run slow on an AMD M 3000+ rig. 3.0+ P4's blast through these things a whole lot faster.

Yes, they are slow in AMD M 3000+ rigs. Most of that is because of the 1.4+ Ghz penalty. I think an AMD M 3000+ is only running at 1.6Ghz (at least mine is). Pentium M still seems to win the points per Ghz for this unit.

 

[DatHak512]

Heh, my newest (for me ) *500MHz* rig just picked up a 600 pointer ...Oh boy, looks like i'll be chewing on this one for a little while...

 

[[Spectre]]

Well my P4 2.8 just picked up p1140_RIBO_FSpeptide_EXT_nospring a 600pointer. Seems to be chewing through it well enough. Nice to get that after no internet connection for a few days last week so I had done work sitting here not going out.

 

[pduan87]

Wow, I can't believe that it took me this long to actually take a peek in this thread. Looks like these big packets give a fat boost in points

Now, looking back in my distributed computing folder, it seems like I am running version 4.00. One question though, I have a dually (2400 MP) and with each of these big work units taking about 100 megs of ram, I have one gig total, will I still be able to play games while running two at a time?

Btw, got my last 2 WU's and then it looks like it's time for an upgrade.

 

[[Spectre]]

The p4 2.8 picked up another 600 pointer..........it got two really small ones in between.

 

[BakedON]

That's funny... my 3.2 did the same thing!
It burped up a couple of teeny tiny WU's in only a few hours and then got ahold of another 600 pointer.....
I hope it has something to do with a particular project. I like to have purpose in my life.

 

[pduan87]

...
Now, looking back in my distributed computing folder, it seems like I am running version 4.00. One question though, I have a dually (2400 MP) and with each of these big work units taking about 100 megs of ram, I have one gig total, will I still be able to play games while running two at a time?
...

Anyone?

 

[marty9876]

Anyone?

My guess is yea. 104x2 + 10 for the console program for a total of ~230 for folding. The OS should fill out things to ~500 meg, leaving 500 meg for a game.

It really depends on the game, but can't hurt anything to try right?

 

[pduan87]

Hmmm... good point. Also, should I let an AMD 1.0ghz tbird run big units?

 

[marty9876]

Hmmm... good point. Also, should I let an AMD 1.0ghz tbird run big units?

It's a personal choice I'd guess. If the boxen is on 24/7, Stanford will benchmark the boxen and won't give it anything it can't handle per ram/cpu specs.

However, waiting ~14 days (just a wild guess) for a WU to drop from a box would bite big time. Just my thoughts...

 

[SmokeRngs]

I'm running the 600 pointers on my system and it only has 512 meg of RAM but doesn't have any problem running games. The two main ones I play are C&C Generals and RTCW. I haven't had any RAM problems yet although it does go into page file usage sometimes.