[jigglylizard]

Would there be any difference in gaming performance in:

400 fsb x6 multi = 2.4 ghz

vs:

266 fsb x 9 multi = 2.4 ghz

I can overclock and reach a certain clock speed either way, but I need higher voltages to get a higher FSB so I'm curious if its actually any better.

Links would be awesome

 

[MrDeodorant]

400x8 = 3.2 GHz, doesn't it? You meant 300x8.

 

[meticadpa]

gaming difference? No. The difference would only be how quickly your RAM runs, which isn't really important, unless you're a Super Pi whore.

Though 400 x 8 would be much faster, since it's 33% faster than 2.4GHz...

 

[jigglylizard]

Quote:

Originally Posted by MrDeodorant 400x8 = 3.2 GHz, doesn't it? You meant 300x8.

I meant 400 x 6 as a comparison.. will edit now

Ninja posts made me look stupid as I can't do basic multiplication today it seems

 

[corky dorkelson]

As a general rule the OC with the higher FSB will be the faster, because obviously the FSB is higher. As far as impact in games goes, only benchmarks will tell you. I suspect the differences will be small, but there should be some difference.

 

[CL3P20]

Quote:

Originally Posted by meticadpa gaming difference? No. The difference would only be how quickly your RAM runs, which isn't really important, unless you're a Super Pi whore. Though 400 x 8 would be much faster, since it's 33% faster than 2.4GHz...

Dont forget, that latency will be a tad worse, once you start lowering the CPU multi.. Pretty sure latency at any speed will effect performance, but it might not be enough to show in-game.

 

[jigglylizard]

Quote:

Originally Posted by CL3P20 Dont forget, that latency will be a tad worse, once you start lowering the CPU multi.. Pretty sure latency at any speed will effect performance, but it might not be enough to show in-game.

So higher multi will mean better/faster latency? Enough to make up for slower FSB/memory speed?

 

[man from atlantis]

500mhzx8 = 4ghz with 1000mhz ram > 445x9 = 4ghz with 1112 mhz.. i can get more points from cinebench r11.5 and dirt 2 benches..

 

[Acoma_Andy]

You could test it yourself, for example look at CPU score in 3DMark '06 at various settings. I know in my case higher FSB and memory speed is faster.

 

[CL3P20]

Im not sure about "making up for it" as you are referring to 2x different latency values..

CPU multi = affecting cache latency for the CPU
FSB/RAM = affecting memory bandwidth and latency

While they both will help for more performance, Im not sure about how you would go about raising 1x enough to negate the affects of the other... but testing should reveal the results you are looking for.

 

[jigglylizard]

Quote:

Originally Posted by Acoma_Andy You could test it yourself, for example look at CPU score in 3DMark '06 at various settings. I know in my case higher FSB and memory speed is faster.

Quote:

Originally Posted by CL3P20 Im not sure about "making up for it" as you are referring to 2x different latency values.. CPU multi = affecting cache latency for the CPU FSB/RAM = affecting memory bandwidth and latency While they both will help for more performance, Im not sure about how you would go about raising 1x enough to negate the affects of the other... but testing should reveal the results you are looking for.

Seems like testing will be the best way to find out. I'll run 3mark06 and 3dmark Vantage, and maybe play Dragon Age for 5 minutes with fraps on to test both settings.

 

[CL3P20]

Please run 'MaxMem' as well with each setup for reporting RAM/FSB latency and bandwidth... you can use the "latency" file from an older version of CPU-Z for reporting CPU cache latency once you have lowered the multi as well...

*I know, latency for CPU cache with my E8400 at stock multi reads:

3ns for L1
15ns for L2

...lowering multi 1x notch.. and:

4ns for L1
16/17ns for L2

So not only am I losing CPU speed, lowering multi, but loosing cache through put and gaining latency.. Your tests will be interesting to look at, if you can match your RAM latency as reported by MaxMem.. and have CPU multi as your only latency change... the affects should be very clear to see.

 

[jigglylizard]

Quote:

Originally Posted by CL3P20 Please run 'MaxMem' as well with each setup for reporting RAM/FSB latency and bandwidth... you can use the "latency" file from an older version of CPU-Z for reporting CPU cache latency once you have lowered the multi as well... *I know, latency for CPU cache with my E8400 at stock multi reads: 3ns for L1 15ns for L2 ...lowering multi 1x notch.. and: 4ns for L1 16/17ns for L2 So not only am I losing CPU speed, lowering multi, but loosing cache through put and gaining latency.. Your tests will be interesting to look at, if you can match your RAM latency as reported by MaxMem.. and have CPU multi as your only latency change... the affects should be very clear to see.

So Maxmem from
http://www.analogx.com/contents/down...m/Freeware.htm

and an older version of CPU-Z? Current version doesn't have that file? (lol @ the downgrade)

 

[CL3P20]

You can DL MaxMem from hwbot as well

...and look for version 1.42-1.48 for CPU-Z to get one that has the latency file with it... its something they got rid of..dont know why..its so damn useful!!

 

[jigglylizard]

I'll get those and run the benchies this weekend. I'll get my OC stable first though and get back to this thread.

 

[ChickenInferno]

I'll just leave these here (I made these about a year ago)

Test setup was EVGa 790i FTW (Analog)
Q6600 on a TRUE w/2x San Ace 120
4gb (2x2gb) Corsair 1600mhz cas9
Corsair TX850
EVGa GTX 260 (Core 192) SC at stock
500gb WD Caviar Black loading games

 

[jigglylizard]

Quote:

Originally Posted by ChickenInferno I'll just leave these here (I made these about a year ago) Test setup was EVGa 790i FTW (Analog) Q6600 on a TRUE w/2x San Ace 120 4gb (2x2gb) Corsair 1600mhz cas9 Corsair TX850 EVGa GTX 260 (Core 192) SC at stock 500gb WD Caviar Black loading games

Awesome! These prove that higher FSB is better! I'll run my own benchmarks just for fun but this is way more data than I was intending to collect, thanks! Even the same CPU

 

[2thAche]

Yes higher FSB is better for games. More bandwidth between CPU and system. Usually there is a point where it gains less and less, and it's a bigger difference say from 266 to 400 than from 400 to 500.

 

[ChickenInferno]

One thing to be noted is that the ram speed was the same in all tests, but the FSB

RAM ratio was very different from the 266mhz and 400mhz (266:800=1:3) vs. (400:800=1:2). I doubt that this difference in ratios acounts for all of the differences, but it would definitly play a role.

 

[jigglylizard]

Quote:

Originally Posted by ChickenInferno One thing to be noted is that the ram speed was the same in all tests, but the FSB RAM ratio was very different from the 266mhz and 400mhz (266:800=1:3) vs. (400:800=1:2). I doubt that this difference in ratios acounts for all of the differences, but it would definitly play a role.

Ok I see. I always heard 1:1 was best however.