865: PAT on/off and FSB overclocking

[Slaimus]

Since turning on PAT reduces the memory access latency, does that mean the max achievable FSB is going to be lower?
Can someone with the Asus or Abit boards post their max stable FSB with PAT on vs off?

 

[Link19]

I just heard that PAT on 865 boards is a rumor. Apparently there is a hacked BIOS that increases RAM speed, which made people believe that PAT could be used on 865 boards.

 

[cmdrdredd]

same thing

 

[Slaimus]

It's not a rumor, it's been proven. Read the Tom's hardware article on it. They actually tell exactly how Asus did it.
I am more interested in whether enabling PAT(reduce memory access latency by 2 clocks) will result in lower max FSB, sort of like running RAM at lower latency will reduce the max stable Mhz it can run at.

 

[Tom]

I actually wondered about the same issue, and thought that a Springdale board might be better for overclocking than a Canterwood because it wouldn't have PAT.

I never found out anything concrete about it though, and since that time there are now "PAT" implementations on Springdale boards, but possibly with the advantage that "PAT" can be turned off, directly, or by using the proper bios. Some day I might try and experiment with it on my IS7.

 

[Slaimus]

Has anyone with a 865 board and the option to enable PAT(hyperpath or whatever each mobo maker calls) tested yet?

 

[cmdrdredd]

Originally posted by: Slaimus
Has anyone with a 865 board and the option to enable PAT(hyperpath or whatever each mobo maker calls) tested yet?

I run 250Mhz FSB on my P4P800 with MAm on (motherboard acceleration mode) and off is similar speed results.

 

[wicktron]

Originally posted by: cmdrdredd

Originally posted by: Slaimus
Has anyone with a 865 board and the option to enable PAT(hyperpath or whatever each mobo maker calls) tested yet?

I run 250Mhz FSB on my P4P800 with MAm on (motherboard acceleration mode) and off is similar speed results.

Question is though, are you running 1:1, or a different ratio? If you run anything other than 1:1, MAM is automatically disabled thus the reason you probably do not see a difference with it ON or OFF.

 

[Ketchup]

Originally posted by: wixt0r

Originally posted by: cmdrdredd

Originally posted by: Slaimus
Has anyone with a 865 board and the option to enable PAT(hyperpath or whatever each mobo maker calls) tested yet?

I run 250Mhz FSB on my P4P800 with MAm on (motherboard acceleration mode) and off is similar speed results.

Question is though, are you running 1:1, or a different ratio? If you run anything other than 1:1, MAM is automatically disabled thus the reason you probably do not see a difference with it ON or OFF.

Actually, I think you only have ro run 1:1 with the 875P. I don't think the limitation is the same on the Pat-enabled 865PE boards. If you would read review Anandtech did, you would see that the Abit IC7 and the PAT-enabled IS7 overclock about the same, and as far as I know, they had PAT enabled.

 

[wicktron]

Actually, I think you only have ro run 1:1 with the 875P. I don't think the limitation is the same on the Pat-enabled 865PE boards.

Quite the opposite. The i875P is PAT enabled at all memory ratio's whereas the i865PE is limited to having pseudo-PAT only at 1:1.

you would see that the Abit IC7 and the PAT-enabled IS7 overclock about the same, and as far as I know, they had PAT enabled.

Getting to a specific overclock point and benchmarks at that point are two very seperate things. The aformentioned 1:1 only limitation for pseudo-PAT on the 865 plays a big role in high FSB, ratio-happy overclocks.

 

[orion7144]

You keep saying pseudo PAT? It's not pseudo PAT it is actualy PAT. The 865 and 875 are the same chipset. Intel just shipped the lower performing 875's as 865. All it took was a bios implimentation to turn it on.

Originally posted by: wixt0r

Actually, I think you only have ro run 1:1 with the 875P. I don't think the limitation is the same on the Pat-enabled 865PE boards.

Quite the opposite. The i875P is PAT enabled at all memory ratio's whereas the i865PE is limited to having pseudo-PAT only at 1:1.

you would see that the Abit IC7 and the PAT-enabled IS7 overclock about the same, and as far as I know, they had PAT enabled.

Getting to a specific overclock point and benchmarks at that point are two very seperate things. The aformentioned 1:1 only limitation for pseudo-PAT on the 865 plays a big role in high FSB, ratio-happy overclocks.

 

[wicktron]

It's pseudo-PAT only for the fact that it is not officially labelled "PAT" by Intel. MAM, Game Accelerator, whatever you want to call it, I'm going to continue to give it the generic name of pseudo-PAT.

 

[cmdrdredd]

well, DDR400 with MAM on is slower than DDR400 at 250Mhz fsb with MAM on which shows that MAM is working as it's the same memory speed

 

[oldfart]

Originally posted by: wixt0r

Actually, I think you only have ro run 1:1 with the 875P. I don't think the limitation is the same on the Pat-enabled 865PE boards.

Quite the opposite. The i875P is PAT enabled at all memory ratio's whereas the i865PE is limited to having pseudo-PAT only at 1:1.

you would see that the Abit IC7 and the PAT-enabled IS7 overclock about the same, and as far as I know, they had PAT enabled.

Getting to a specific overclock point and benchmarks at that point are two very seperate things. The aformentioned 1:1 only limitation for pseudo-PAT on the 865 plays a big role in high FSB, ratio-happy overclocks.

PAT (MAM, GAM, etc) is normally off on all 875 and 865 boards when using a ratio. Once Asus figured out how to enable it on the 865 chip, things changed. If the mobo mfgr decides to make a BIOS that will do it, PAT can be enabled on the 875 or 865 when using a 5:4 ratio. CPUz version 1.18c will test for PAT disabled/enabled.

 

[Slammy1]

You need to run MAM AND set performance mode to Turbo on the P4P800 to get PAT. MAM seems to give improvement at all dividers for me, but if I also run Turbo it is not stable. One or the other (not both).

 

[cmdrdredd]

[qPAT (MAM, GAM, etc) is normally off on all 875 and 865 boards when using a ratio. Once Asus figured out how to enable it on the 865 chip, things changed. If the mobo mfgr decides to make a BIOS that will do it, PAT can be enabled on the 875 or 865 when using a 5:4 ratio. CPUz version 1.18c will test for PAT disabled/enabled.[/quote]

where does it show if PAT is enabled or not?

 

[oldfart]

You need the 1.18c version. Under the memory tab, performance mode is enabled or disabled.

 

[THUGSROOK]

on i875 (and some i865) boards PAT / MAM should be enabled using the latest bios ~ even in ratio modes.

 

[cmdrdredd]

I don't see 1.18c on the main site anywhere. got a link?

 

[oldfart]

1.18c here

 

[cmdrdredd]

TY

 

[thermite88]

Originally posted by: oldfart
PAT (MAM, GAM, etc) is normally off on all 875 and 865 boards when using a ratio. Once Asus figured out how to enable it on the 865 chip, things changed. If the mobo mfgr decides to make a BIOS that will do it, PAT can be enabled on the 875 or 865 when using a 5:4 ratio. CPUz version 1.18c will test for PAT disabled/enabled.

I was using the 1.008 BIOS. I had the MAM and TURBO enabled. I tried both 5:4 and 1:1 ratio, but CPU-Z said PAT was off no matter what I did. Flashed it to 1.009 BETA from the German site. Same result.

CPU: 2.6C OEM with Zalman 7000-CuAl cooler
MEM: 2x512MB PC3200 from OC System (with PC2700 SPD)
MB: Asus P4P800 Deluxe
Video: Sapphire 9100 128MB Pro
Storage: LSI MegaRaid Express 500 with 4 Ultra160 10K drives in Raid-5

I was able to OC to 238FSB stable using:
Vcore: 1.675
DDR: 2.75

At 240FSB, system runs normally, but lockup running SANDRA torture test.

 

[thermite88]

Any comments on the post at ABXZone on how to turn on PAT when overclocking?