Abit IX38 Quad GT - Living Review II

[supershanks]

welcome back vwgolfman,
Nice to see you making progress. As I tjink we mentioned earlier small ffts stresses the cpu . Large ffts memory. Blend runs each alternatively so testing the whole. With the same logic your cpu temps will be lower using blend than small ffts. Using the resukts from this poat talkig about your Zalman CNPS9700-LED Super Aero Flower Cooler

Applying your fft result your max will be 65C hit on small fft runs, but on blend your average wil be ~5C lowe maybe 59~60C you'll need to note them on your blend test. So Yep I'd be pretty happy running those temps , though it's your decision of course

If you can run your memory at is stable at 4-4-4-12 @ 852 1.8V fine stick with it, thats good, I find with mine it still needs about 2v at that speed. I think 2.15v is within the voltage envelope for your ram, so your ok, but like everything less is morere , so if you don't need the voltage don't use it, you get less heat as a benefit.

It is that Corsair Dominator stuff with the heatsinks so will I be OK temperature-wise?

I'm not quiet sure what you mean here ?? But if you run below 2v your memory should be ok without active cooling , if that's your question.

luck

[BUFF]

Originally Posted by karas

just a question.. on my ix38qgt the multiplier for ram PC2-8500 is "1:2.00 (DDR2-1088)" ? why 1088 ? *_*

abits typically run a ~2% overclock out of the box.
I think that if you check you'll find your CPU bus ~2% over the CPU's rated stock.

[karas]

Originally Posted by BUFF

abits typically run a ~2% overclock out of the box.
I think that if you check you'll find your CPU bus ~2% over the CPU's rated stock.

loool it's true xD thanks

EDIT: trying to overclock a little the CPU at 1:1.00.. 300fsb.. with the Small FFT test cpu temp = 45°C but what about the "PWM Phase 0,1,2..." ? What are they ? The temp seen on uGuru is correct for them ? (with this test it's ~70-75°C O_o)

[supershanks]

PWM are the digital circuits that replace the mossfets on earlier boards. It stands for Pulse Width Modulation. It's used to provide power regulation & stable current, it works by switching power on & off, ie a pulse.
CPU VTT voltage is the main factor controlling PWM. It can be used in conjunction with Vcore & GTL ref, to provide a stable platform for the CPU.

[karas]

mhh.. i see.. and what about the temperature ? with blend it goes at 70-75°C stable.. is it bad ? :/

[supershanks]

70~75 is fine, have a look at some of the screenshoots of Guru in the benchmarks etc., linked in Post' #1

[supershanks]

New beta bios Courtesy of Gilgamesh M704B_12.09
There is an changelog in this but i can't say i've learnt anything from it. If any body has any ideas please let me know.
WARNING I was sloppy when i updated, didn't load defaults before flashing, flashed from 8x470. As a result I had C1 error. Only managed to cure it by using cheapo ddr2 stick & changing voltage So don't do as I do. Flash at your own risk.




M704B_12.B09 - Initial Feedback

[supershanks]

1) Pi & Everest Benches Comparison 8x470@1:1.20
________32M PI_______1M PI_________WP32M_____WP 1024_______Everest R_____ W______COPY___Latency
12.05____14.22.312____13.656________12.265______39 0.296_______9210________8437____9069____56.3
12.06____13:53.766____13.672________11.675______38 9.593_______9204________8463 ___9094____56.4
12.08____14.06.907____13.640________11.922______38 8.343_______9279________8509____9136____56.0
12.09____14:03.750____13.672________11.437______35 8.078_______9269________8493____9131____56.2
12.09 8x470@1:1:20 Benches


12.08 8x470@1:1:20 Benches



Aquamark
_____Aquamark___CPU_____GFX
12.06 201,845___21,450___38,112
12.08 202,088___21,490___38,160
12.09 202,591___21,357___38,523

12.09 8x470@1:1:20 Aquamark


12.08 8x470@1:1:20 Aquamark


For Ethel
AND 1:1.25 Works
12.09 8x470@1:1:25 1175Mz
Went direct to :1.25 on previous settings , Post Froze. But there is good news , the OC recovery cut in & smoothly took me to previous settings. Every other attempt on previous BTAs has needed at least a clear Cmos. Slackened tras from 10 to 12 & got there , benches don't look much better that Will rerun later with old timings & voltage tweaked

________32M PI_______1M PI_________WP32M_____WP 1024_______Everest R_____ W______COPY___Latency
1:1.20____14:03.750____13.672________11.437______3 58.078_______9269________8493____9131____56.2
1:1.25____14:19.891____13.687________11.344______3 61.780_______9249________8507____9131____56.3

Update: - Following reboot got C1 Error. Usual resolution, clear cmos, cheapo ddr2 stick vdimm to 2v & back up. Clocked down to 450 x 8 @1:1.25 1125Mz & got the same.

Seems a definite improvement but not there yet, may need a bit of tweaking, but you tend to lose enthusiasm doing repeated C1 Recoveries.


3DMark06 12703

Guru CPU Temp
Just noticed last night that the Guru Temp, corrected in an earlier BTA is now back to reading nonsense again.

[glowball]

Hmmm, and there was you hoping the next one was probably going to be a 'final'.
Still, mustn't complain, at least they are getting updated regularly.
Guess the 1:1.25 is still a problem, personally I'd like to see something between that and 1:1.5.

EDIT: Should have read comments on xtremesystems first, it appears 1:1.25 is now working so I will go ahead and try this one!
Well done Abit (if this correct).

[supershanks]

Hi m8,
well

nd there was you hoping the next one was probably going to be a 'final'.

wasn't particularly hoping , just thought it might be there.
Have done benches but not tried 1:1.25 yet after the c1 incident when i flashed

will try it after.