Request: FX series Single-Thread

[vampirr]

Hello, I have a request for people that own any Bulldozer/Piledriver CPU...

So I think we can't see really a true Single-Thread performance of AMD's Bulldozer/Piledriver CPU's, why?

Because by default Cool n Quite feature is ON in BIOS so I would like to see comparison between single-thread performance with and without Cool n Quite feature to see if it affects its single-thread performance in programs, older games and newer ones...

If someone can do the tests, I hope you will do it and post your results...

 

[tweakboy]

The FX series is much slower in single threaded apps then a Sandy or Ivy. gl

 

[Maximilian]

Its cool & quiet* and tbh i think something like C&Q affecting performance would have been discovered long before now, these chips have been out for ages. If people can pinpoint the windows scheduler as a potential problem (just after BD's release) then i doubt they would have overlooked a simple on/off option in the BIOS.

But by all means go wild, do these unspecified "tests" and see if it makes a difference with anything I would be very surprised if it did.

 

[blckgrffn]

Turbo seems to work so well with the newer AMD chips, I can't imagine what we'd find.

Under load, I've observed them to just peg out at their Turbo frequencies for hours/days at a time. If C&Q was at work you'd expect something else.

 

[vampirr]

I don't own any AMD's Bulldozer of Piledriver series, if anyone has one please do a test or something. I am curious, also when people say that new AMD's CPU's perform badly on older games like 3+ years then maybe cool & quite is maybe the cause of it.

 

[Bearach]

Hello, I have a request for people that own any Bulldozer/Piledriver CPU...

So I think we can't see really a true Single-Thread performance of AMD's Bulldozer/Piledriver CPU's, why?

Because by default Cool n Quite feature is ON in BIOS so I would like to see comparison between single-thread performance with and without Cool n Quite feature to see if it affects its single-thread performance in programs, older games and newer ones...

If someone can do the tests, I hope you will do it and post your results...

Cool 'n' Quiet may have a very minor performance hit in getting back up to full speed - barely noticeable if at all - but I don't think it will sway the numbers much. Also remember, Intel, has SpeedStep which does essentially the same thing.

However, I have a FX 8350 based system, and would be happy to do some tests for you. What kind of tests are you hoping for?

NOTE: My system is stock everything.

 

[Bearach]

Turbo seems to work so well with the newer AMD chips, I can't imagine what we'd find.

Under load, I've observed them to just peg out at their Turbo frequencies for hours/days at a time. If C&Q was at work you'd expect something else.

I've noticed this too, with my system. Even under very heavy load, 2 modules tend to turbo up to 4.2GHz.

Some exceptions however is video encoding, Prime95, and LinX.

 

[Durvelle27]

I also have an FX 8320 system and would be happy to do some test

 

[vampirr]

Cool 'n' Quiet may have a very minor performance hit in getting back up to full speed - barely noticeable if at all - but I don't think it will sway the numbers much. Also remember, Intel, has SpeedStep which does essentially the same thing.

However, I have a FX 8350 based system, and would be happy to do some tests for you. What kind of tests are you hoping for?

NOTE: My system is stock everything.

Test this games with and without "cool & quite"...

Freeware:
Warzone 2100 (RTS, CPU Intensive when alot of units in multiplayer)
UFO: Alien Invasion (turn based, modified Quake 2 engine)
Cube 2 (indie game, first person shooter)

I really don't know what games do you own but if you have these games:
Homeworld/Homeworld 2/Supreme Commander

Quake 1-2-3/Half Life/Red Faction/Far Cry/ARMA

Need For Speed: Porsche/ Need For Speed: Underground 1-2/ Flatout 1-2

So do a test with those games, if you have old games from 1996(start of 3D game era) to Q3 2011(slightly before FX series release)...

 

[Bearach]

Test this games with and without "cool & quite"...

Freeware:
Warzone 2100 (RTS, CPU Intensive when alot of units in multiplayer)
UFO: Alien Invasion (turn based, modified Quake 2 engine)
Cube 2 (indie game, first person shooter)

I really don't know what games do you own but if you have these games:
Homeworld/Homeworld 2/Supreme Commander

Quake 1-2-3/Half Life/Red Faction/Far Cry/ARMA

Need For Speed: Porsche/ Need For Speed: Underground 1-2/ Flatout 1-2

So do a test with those games, if you have old games from 1996(start of 3D game era) to Q3 2011(slightly before FX series release)...

Okay, but I think with a majority of those I'm going to be GPU limited.

These tests are to be done VSYNC OFF? Do these freeware games have any benchmark mode? What is your testing method?

EDIT: I really don't think the 8350 will break a sweat running most of these.

EDIT2: Wouldn't CPU tests be any better?

 

[vampirr]

Okay, but I think with a majority of those I'm going to be GPU limited.

These tests are to be done VSYNC OFF? Do these freeware games have any benchmark mode? What is your testing method?

EDIT: I really don't think the 8350 will break a sweat running most of these.

These freeware games does not have a benchmark mode, when playing Warzone 2100 just find a room with 4 or even 8 players if possible because mostly there are alot of buildings in each base 20-30+ and most players have 100+ units or even 250 max and pathfinding is calculated by your CPU.

There isnt actually any method, when playing UFO:Alien Invasion just make a save at the beginning of each level and then play it with "cool & quite" ON and then Off the whole round.

Cube 2 has a non-standard engine, so do multiplayer... It wont give you very accurate results but do a stress test with large amounts of players...

My point it test the pre 2011 games with and without cool & quite, use your methods that you think its fit for each game. The point is that I want to see if without cool & quite you have better single threaded performance in older games than with cool & quite... Its simple.

Thank you for doing the tests in advance.

-edit- do any other tests that you want, I want to see results in older/pre 2011 games. The older the better, maybe.

 

[Durvelle27]

3Dmark05 & 3DMark06 are CPU limited also

 

[KingFatty]

Maybe you can think about the mechanism of how cool n quiet works, and that will help put to rest any concerns that will be verified by the results.

Essentially it allows a chip to slow down and use less power when it's idle.
http://en.wikipedia.org/wiki/Cool'n'Quiet

But when the chip is being used, it's not hampered at all, so I'm not sure what mechanism you think might be enabling the chip to be affected? Under testing conditions the chip will not be idle, so cool n quiet will not be affecting the chip whether it's set in the BIOS to be on or off? Disabling it will just make you use more power when the chip is idle?

 

[nenforcer]

None of those older games are multithreaded or multi-core aware. They come from a single core era. You are probably better off disabling 3 out of the 4 modules, and 7 out of the 8 "cores", leaving only a single integer unit and 1/2 FPU available for testing single threaded performance.

 

[Jovec]

8350/5800 single core performance is about on par with the cheapest, slowest Ivy Bridge (G1610) single core. Yes, I have all three.

What gets lost in this argument is that one manufacturer is running their CPUs essentially as fast as they can (AMD) while the other is leaving performance on the table (Intel).

 

[Bearach]

3Dmark05 & 3DMark06 are CPU limited also

Ran 3DMark05, found difference to be within the margins of error. There has been so far no "real" difference with it on, or off.

I can upload videos of both runs for the OP to see, if he wishes.

CnQ OFF

3DMark Score: 23190 3DMarks
GT1: 80.707839966 FPS
GT2: 64.27684021 FPS
GT3: 153.859298706 FPS
CPU Score 25201 CPUMarks
CPU Test 1: 15.923458099 FPS
CPU Test 2: 17.726837158 FPS

CnQ ON

3DMark Score: 23182 3DMarks
GT1: 80.583244324 FPS
GT2: 64.075065613 FPS
GT3: 154.411865234 FPS
CPU Score 25012 CPUMarks
CPU Test 1: 15.947370529 FPS
CPU Test 2: 17.434801102 FPS

 

[Rvenger]

Ran 3DMark05, found difference to be within the margins of error. There has been so far no "real" difference with it on, or off.

I can upload videos of both runs for the OP to see, if he wishes.

CnQ OFF

3DMark Score: 23190 3DMarks
GT1: 80.707839966 FPS
GT2: 64.27684021 FPS
GT3: 153.859298706 FPS
CPU Score 25201 CPUMarks
CPU Test 1: 15.923458099 FPS
CPU Test 2: 17.726837158 FPS

CnQ ON

3DMark Score: 23182 3DMarks
GT1: 80.583244324 FPS
GT2: 64.075065613 FPS
GT3: 154.411865234 FPS
CPU Score 25012 CPUMarks
CPU Test 1: 15.947370529 FPS
CPU Test 2: 17.434801102 FPS

.0000002% difference... margin of error.

 

[Bearach]

.0000002% difference... margin of error.

Yeah, which is what I said.

Note: My graphics card is pretty old, and in need of replacement to match the processor better. AMD 6850.

Also tried to run Warzone but found it hard to find a server, and one that was ready... So couldn't test really.

UFO Alien, seemed to be stuck for me, always at 44fps FRAPs reported to me, and that didn't change with the CnQ off. Ran with default settings.

Will continue some more tests when I have some more time, but so far I think there is little evidence to me, that CnQ, is a cause of performance issues with older games.

 

[Durvelle27]

Ran 3DMark05, found difference to be within the margins of error. There has been so far no "real" difference with it on, or off.

I can upload videos of both runs for the OP to see, if he wishes.

CnQ OFF

3DMark Score: 23190 3DMarks
GT1: 80.707839966 FPS
GT2: 64.27684021 FPS
GT3: 153.859298706 FPS
CPU Score 25201 CPUMarks
CPU Test 1: 15.923458099 FPS
CPU Test 2: 17.726837158 FPS

CnQ ON

3DMark Score: 23182 3DMarks
GT1: 80.583244324 FPS
GT2: 64.075065613 FPS
GT3: 154.411865234 FPS
CPU Score 25012 CPUMarks
CPU Test 1: 15.947370529 FPS
CPU Test 2: 17.434801102 FPS

CnQ Off

3DMark Score: 9083 3DMarks
GT1: 44.802448273 FPS
GT2: 25.093877792 FPS
GT3: 42.659263611 FPS
CPU Score: 29187 CPUMarks
CPU Test 1: 17.72958374 FPS
CPU Test 2: 21.354335785 FPS


Edit: Ignore the GPU scores as i was using a GT 440 2GB

 

[Bearach]

CnQ Off

3DMark Score: 9083 3DMarks
GT1: 44.802448273 FPS
GT2: 25.093877792 FPS
GT3: 42.659263611 FPS
CPU Score: 29187 CPUMarks
CPU Test 1: 17.72958374 FPS
CPU Test 2: 21.354335785 FPS


Edit: Ignore the GPU scores as i was using a GT 440 2GB

Tried to have task manager on second monitor - All the constant resolution changes made it a pain - but from what I can see 3DMark really was poor at utilising my processor.

 

[Durvelle27]

Tried to have task manager on second monitor - All the constant resolution changes made it a pain - but from what I can see 3DMark really was poor at utilising my processor.

because it can't utilize more than dual core as its mostly single threaded

 

[vampirr]

That are just test with a benchmarking tool(synthetic), I asked that you guys test in older games with c&q on and off. Anyway those tests are wellcome, if someone can knows how to get most out of single-thread performance from stock clocked FX processors then bring it.

I am just curious... Thanks for doing some tests and benches, big love to you all. No homo.

 

[Durvelle27]

That are just test with a benchmarking tool(synthetic), I asked that you guys test in older games with c&q on and off. Anyway those tests are wellcome, if someone can knows how to get most out of single-thread performance from stock clocked FX processors then bring it.

I am just curious... Thanks for doing some tests and benches, big love to you all. No homo.

Running Need For Speed Underground 2

 

[Bearach]

That are just test with a benchmarking tool(synthetic), I asked that you guys test in older games with c&q on and off. Anyway those tests are wellcome, if someone can knows how to get most out of single-thread performance from stock clocked FX processors then bring it.

I am just curious... Thanks for doing some tests and benches, big love to you all. No homo.

Yeah, I've given some feedback to some games also. I've tried running my copy of Homeworld 2... Wouldn't run, guessing Windows 7/8 has problems with it?

Going to run some old Need For Speeds, Far Cry, and some others too.

 

[vampirr]

If you can't run some games, try to run it with compatibility mode or find some fixes.

Search the some forums, or if you can then install Windows XP if you have a legit copy.