As promised. PIII-S data

[Mictlan]

Well, here is all the data I currently have for my system.

Complete hardware:

PIII-S 1.26 GHz chip with 512 Kb of L2 memory.
Iwill DVD266u-RN mobo, with VIA Apollo 266T chipset.
2 sticks of 256 Mb of local brand of DDR memory, with Micron 7.5 ns chips.

Here is the WCPUID data at stock speed. Seems that Iwill is a little off with the Microstepping FSB, as in the BIOS I have 133 FSB.

 

[Cy]

Pretty sweet...how does she crunch?

Cy

 

[Mictlan]

As you can see, it does not display DDR in the memory. That seems odd to me. I'll have to ask someone else using the PIII-DDR combo to shown in WCPUID. Maybe WCPUID check in the processor side of the memory controller. Really don't have a clue.

Anyway, Sandra information of the chip.

 

[Mictlan]

Sandra CPU benchmark.

 

[Mictlan]

Sandra memory scores. This is using the latest Sandra software.

As you can see the scores are comparable to the ones shown with Via Apollo controller. I left out the 850 chipset as we all know that even a dual pumped memory controller is no match to the quad pumped one.

 

[Mictlan]

Finally, here is an extract of the SETISpy log for the last WU crunched in my system. I leave you to draw the conclusions of them .

 

[Cy]

Looks pretty reasonable to me. Just that question about the DDR memory not showing in Sandra...

Cy

 

[Morpheus]

not sure what's up with that DDR, Juan.... I do know the P3 really can't take advantage of it....

When I get home I'll post mine PIII-S (with Mushkin PC-150), I think I have gotten down MID to LOW 3s for WU average now (3.2-3.5)... But I am still teaking with it...

 

[Mictlan]

Kris:

The 3.2-3.5 rangle to 0.4xx AR WU is overclocking the chip right? If you have some result running at stock (133 FSB) can you post them to get a better comparison.

I'm running 3.8 with 0.4xx AR WU running at stock (well, almost) Given some time, I'll see how far can I get this little chip, maybe tonight I'll try reaching 150 FSB . ASAP I'll post my results.

 

[Mictlan]

not sure what's up with that DDR, Juan.... I do know the P3 really can't take advantage of it....

When I get home I'll post mine PIII-S (with Mushkin PC-150), I think I have gotten down MID to LOW 3s for WU average now (3.2-3.5)... But I am still teaking with it...

I've heard the DDR + PIII setup is not that efficient. It's like using DDR333 with K7 processors. The available memory bandwidth is higher than the processor bandwidth. If someone want to read a little more about this you can here. Even with this against, as I was looking for a dual PIII mobo that supported the new Tualatin core, I ended buying the DVD266u, that supports two Tualatins processors and DDR. As an added bonus I'll take the 3-6% increase in processing capacity

 

[Yodums]

Looks weird Juan. Especially the Sandra posting memory scores that low.

Although the times look great.

 

[Mictlan]

Especially the Sandra posting memory scores that low.

Why do you think that the Sandra Memory scores are low? We are talking about a chip that don't have the bus capacity to saturate the memory. If you look in the link I posted above, the PIII have a 1.06 GB/s processor bus bandwidth. A DDR memory bus bandwidth using 133 MHz will be 2.1 GB/s, so I won't be seing a K7 Sandra Memory score, much like a K7 + SDRAM one.

That's why in a dual setup using a PIII it will pay to use a DDR memory controller. PIII dual architecture uses a common memory controller, so each request is procesed by the same controller, not like the AMD system that uses independent memory controllers for each chip. So 1.06*2=2.12 GB/s, just enough to saturate the controller.

 

[[P3D]AndrewM]

Finally, here is an extract of the SETISpy log for the last WU crunched in my system. I leave you to draw the conclusions of them .

Well , the times seem like on a 1.3G duron with DDR . But the price difference is ....hmm , a bit higher...

 

[lonewolf1983]

my 1.2 duron is running at 1.3 and turning out times above 4 hours
probably due to the 100 (110 overclocked) bus and less cache

those are some good times

 

[BadThad]

Nice SETI times! For the 417 angle ranges my system (in sig) takes 3:10 to complete a wu...and I have a way higher bandwidth. Obviously, it's the cpu that's doing a fine job at crunching.

 

[Mictlan]

I want to go back to the Sandra Memory scores. I'm using the latest Sandra (I think it is 2002.1.8.59) and somewhere I read that you'll have to compare SAndra scores only against the same version.

Also, this is running the chip at stock (1.26 GHz), I'm just waiting for the weekend to oc this chip and see if I can get 3:15 WU or less.....

 

[Yodums]

Mictlan:

Since it won't take advantage of DDR it'll still have equivilant SDR scores, and the old 440BX chipset scored 800/800 on average, the recent ones scored around 1000...

Yodums

 

[Mictlan]

Since it won't take advantage of DDR it'll still have equivilant SDR scores, and the old 440BX chipset scored 800/800 on average, the recent ones scored around 1000...

Please check your Sandra scores, they might be for the SiSoft Sandra 2001 benchmark, if you see in my screenshot above, the 440BX has a lower score (415/475).

In there is also the VIA Apollo 266 (340/472) and the VIA KT266 (490/650), just to show some of them.