Benchmarks 8K7A Mem @183MHz FSB

[TheTeacher]

Ok, I have begun my benchmarks. Mem timings at 8-8-8-3-2-2.5-2
822/999 FSB on the first try, usually second gives it a healthy boost. More to come. CPU is at 1.466GHz
Mike

 

[chainbolt]

thanx, as I said, non-linear impact, I have seen benchmarks with the A7M266 at FSB 170 and a score of 1050. I get Sandra 730/930 with 157Mhz. Above 160MHz the impact on bandwidth is dramatically shrinking. I think you have to find the best mutliplier/FSB Mhz combination for the highest bandwidth, that is probably lower than 185Mhz.

Please post more benchmarks! This is REALLY interesting.

 

[blackhawk]

How do you get 183 fsb? My Epox only goes to 166 but I run my duron at 160.

8-8-8-2-2-2-2

714/913 on Sisoft mem bench

 

[Doomguy]

chainbolt: How come in your 3-way comparison you used the new slower bios on the epox? Just curious to know why.

 

[TheTeacher]

With memory, you can only get a certain bandwidth out of it. mine is 1019. Even if it is a higher speed than that, it will actually slow down. I am not sure why this is. It is all dependant on the memory. SO right now my memory is the bottleneck. Nothing I can do about it. Using the same timings at 166 to get 1019, I raise the FSB to 175MHz, and it actually is slower, even with a higher clocked CPU it is slower. I believe it has to do with noise. Not sure though. I will learn more about this later.
Mike

 

[TheTeacher]

Using latest BIOS from epox, 5/21/01 at 183 I get 768/1000
A decrease of 54MB/s. and a gain of... 1MB/s... LoL HOLY 1 MB/S!!!! AAHHHHH =P

Lets see how this new BIOS handles 190MHz FSB. LoL = )
Mike

 

[kzyazn]

where are you finding the bios for this? i looked at the list and i don't see EP-8K7A

 

[TheTeacher]

I myslef got it directly from epox, but you can find it at various websites. Ok listen to this one guys. When I try the mem timings at 8-8-8-2-2-2-2 at 166FSB I get 503/941 Why are they lower then when it is at 8-8-8-3-2-2-2? Shouldn't it be higher due to lower timings? Man, I really need to find out what effect these settings have on RAM
Mike

 

[TheTeacher]

Ok, to verify, I run the test again. This time I get: 760/992. That sounds more realistic. Now, why does it go so low the first time? It has done it the last three times I have used these settings.

 

[TuffGuy]

<< When I try the mem timings at 8-8-8-2-2-2-2 at 166FSB I get 503/941 Why are they lower then when it is at 8-8-8-3-2-2-2? Shouldn't it be higher due to lower timings? >>


no, cas2 is faster than cas3 when it comes to latency. it's not like with the fsb, where a higher number is better.

a lot of sticks of ram will do 150mhz+ cas 2, but they will only hit 160mhz+ at cas3.

 

[TheTeacher]

Hey guy... This is DDR, there is no CAS3 there is 2.5 That is not CAS latency that is another setting. But, I know what you mean.
BTW 174.4MHz @ CAS2 790/1030
Mike

 

[Renob]

What brand of ram are you using

 

[TheTeacher]

OOps I typed in those numbers wrong... it is LOWER at 8-8-7-2-2-2-2 it is HIGHER at 8-8-8-2-2-2-2

*EDIT*
PC2400 CAS2 Corsair 256MB Highest at CAS2 is 174.5MHz, highest at CAS2.5 is 190MHz.

 

[TuffGuy]

like i said before. ddr, sdram or whatever. shorter latency improves performance.

 

[TheTeacher]

It is opposite what you said. Read it slowly.
Mike
*I don't mean to sound rude. I just haven't slept in a few days*

 

[TuffGuy]

right. it's quicker/higher at 7 than it is at 8.

 

[Renob]

xSuicidalx can you change the fsb and clock multiplier in the bios???

 

[TheTeacher]

Ok, so I typed it wrong... never mind... LoL it is LOWER!! = P At 8-8-7-2-2-2-2 and HIGHER at 8-8-8-2-2-2-2
=P i need sleep maybe.
Mike

 

[chainbolt]

<<

<< When I try the mem timings at 8-8-8-2-2-2-2 at 166FSB I get 503/941 Why are they lower then when it is at 8-8-8-3-2-2-2? Shouldn't it be higher due to lower timings? >>


no, cas2 is faster than cas3 when it comes to latency. it's not like with the fsb, where a higher number is better.

a lot of sticks of ram will do 150mhz+ cas 2, but they will only hit 160mhz+ at cas3. >>




CAS 3? DDR mobos don't have CAS 3. DDR comes only as 2.5 and 2.

 

[TheTeacher]

I ALREADY SAID THAT!!!
Mike

 

[chainbolt]

<< I myslef got it directly from epox, but you can find it at various websites. Ok listen to this one guys. When I try the mem timings at 8-8-8-2-2-2-2 at 166FSB I get 503/941 Why are they lower then when it is at 8-8-8-3-2-2-2? Shouldn't it be higher due to lower timings? Man, I really need to find out what effect these settings have on RAM
Mike >>



Thanx again, it's difficult to tell what is the impact of the board and what is the impact of the memory. You can get with good memory, Sandra scores of up to 1000 with FSB 170. I got 704/930 with 157. So, I guess your scores would be higher (they should be higher) with better memory. Did you optimize anaything? Taking out all PCI cards and unplugging all USB devices. Sandra is just a little toy, with some bugs (USB!). We shouldn't take this too serious. There is another bandwidth test &quot;memtach&quot;

 

[chainbolt]

<< I ALREADY SAID THAT!!!
Mike

*EDIT*
That setting is NOT CAS LATENCY DUE TO POPULAR RETARD BELIEF!!! I still need sleep. heh >>





So, what is it then? I thought it's the latency setting, the most important one among the AMD 761 memory settings.

 

[TheTeacher]

Ahhh.. Sorry for that out burst... I am really PISSED off at my cable modem from CHARTER WHO SUCKS... i will stop there... It is not CAS latency. It some some other latency, it has a BIG effect on performance when other settings are at certain numbers. If it is 8-8-7-2-2-2-2 it has no effect except on stability, when it is at 8-8-7-2-2-2-3 then I change it to 8-8-8-2-2-2-3 it has a big hit on performance and very little on stability. Right now, 180MHz, 8-8-7-3-2-2.5-2

 

[TheTeacher]

Ok new benchmark at 175 8-8-8-2-2-2-2 coming up in a flash!
Mike
*EDIT*
802/1019
Arrghhh... I had it 820/1033 earlier, and that was CAS2.5!!!

 

[chainbolt]

ahhh, congratulation, you broke the barrier! And it again confirms, the highest FSB frequecny DOES not give the highest bandwidth, I think it's according to one of these base laws from physics: I don't know how to put in English: something like diminishing impact when approaching the limit, or so....