Overclocking Sandbox: Tbred B DLT3C 1700+ and Beyond

PSU intake fan location

Recently, I got a Fortron 530W. When I ordered the PSU, I did not realize its unusual fan location compared to many other PSU. The fan intake location is at the front and NOT at the bottom of the PSU.

I found that it did not cool the CPU surrounding as good, few degree (3-4) C higher measured, as a PSU with bottom fan intake. I consider this as an important factor in choosing PSU.

I then look up the AMD technical link and found an AMD tech doc recommends to have PSU fan intake above CPU area, at the bottom of PSU:

QUOTE:
"Power supply with bottom intake"
"Intake only in the front is not optimal"
"Bottom intake cools the CPU best"

Reference
AMD Athlon System Cooling Guidelines

If only fortran/sparkle had a nice 550w PS with that 120mm fan underneath like their 300w & 350w units.

This is a summary of the indiividual results posted before.

Comparing some HSF's on CPU overclock frequency on air

Reference HSF:
HSF 0.23 C/W e.g. SLK-800(U)
Tbred B 1700+ DLT3C 2500 MHz, 1.9 V, 50 C load
5 C rise in ambient, lower CPU overclock frequency by 5.3%

HSF 0.25 C/W e.g. SK7
CPU overclocks 30 MHz lower compared to SLK-800(U).
5 C rise in ambient, lower CPU overclock frequency by another 5-6%

HSF 0.30 C/W e.g. Vantec Aeroflow, Volcano 7+
CPU overclocks 212-300 MHz lower compared to SLK-800(U), to 2200-2288 MHz
5 C rise in ambient, lower CPU overclock frequency by another 6.7%

HSF 0.36 C/W e.g. Volcano9, Volcano 11
CPU overclocks 347-433 MHz lower compared to SLK-800(U), to 2067-2153 MHz
5 C rise in ambient, lower CPU overclock frequency by another 6.7%

How are the details about how above numbers were estimated:
Reference:

Impact of higher ambient temperature on CPU clock frequency (page 6)

Comparing different HSF's and impact of ambient temperature on max CPU frequency (HSF 0.3 C/W) (page 6)

Comparing different HSF's and impact of ambient temperature on max CPU frequency (HSF 0.36 C/W) (page 7)

of Tbred B 1700+ DLT3C 2.5+ GHz air results

So, Me an my NF7-S... what a concept... after experiencing the infamous BIOS blackout after a reboot... my emergency kit came in, and I did exactly by the book on what was to do per instructions from excaliber and another member here in our forums... Audioaficionado... (a million thanks to you bro)

Now, get this... I couldn't get my xp1700 0319 up passed 207FSB... now after reprogramming the zapped bios, I'm running 220FSB @11.5multi... vcore v1.92, vdmm v2.8, v1.7on chipset, temps idle/load: 40.5c/44.5c Heatsink:SLK-900u w/tornado 92mm 119cfm. CorsairXMS lowlatency 2x256mb... timing:2.5-2-2-2-5

cpuid
Processor #1 : AMD Athlon XP (Model 8) / 7C4F5C41
Platform : Socket A (Socket 462)
Vendor String : AuthenticAMD
CPU Type : Original OEM Processor (0)
Family : 6 (7)
Model : 8 (8)
Stepping ID : 1 (1)
Name String : AMD Athlon(tm) XP

Internal Clock : 2535.75 MHz
System Bus : 441.00 MHz DDR
System Clock : 220.50 MHz
Multiplier : 11.5

L1 I-Cache : 64K Byte
L1 D-Cache : 64K Byte
L2 Cache : 256K Byte
L2 Speed : 2535.75 MHz (Full)
Host Bridge : 10DE:01E0.C1 [NVIDIA nForce2]
South Bridge : 10DE:01EB.C1 [NVIDIA nForce2 MCP]
VGA Device : 10DE:0152.A4 [NVIDIA GeForce2 Ultra]
Memory Size : 512M Byte
OS Version : Windows XP Version 5.01.2600 Service Pack 1


##--- Date 06/25/2003, Time 22:37:45

Folding continues to work in the background while i'm posting here.
Go figure on the zapping of the bios an reprogramming it to get the extra boost. who knew?

9500np to 9700 Softmod and 3D Mark w/ Tbred B 1700+ DLT3C

Got a 9500 np, added the catalyst 3.4 9700 hacked driver and it seems the mod is successful.

The successful rate is only between 1/3 to 2/3 (risky), so the 9700np (~$200) is better choice, if probability of success is taken into account.

CPU overclocked to 2534 MHz (203 x 12.5)

3DMark03 before mod (default) = 2897
3DMark03 before mod (340/300) = 3542
3DMark03 before mod (369/302) = 3737
3DMark03 after mod (277/270 default) = 4519
3DMark03 after mod (331/306) = 5214
3DMark03 after mod (342/311) = 5304 (update)
3DMark03 after mod (351/308) = 5379 (update)
3DMark03 after mod (360/308) = 5465 (update)
CPUMark 639.0, Fill Rate (Single-Texturing) 1578.8 MTexels/s, Fill Rate (Multi-Texturing) 2505.5 MTexels/s
3DMark03 after mod (360/311) = 5487 (update)
CPUMark 637.0, Fill Rate (Single-Texturing) 1587.5 MTexels/s, Fill Rate (Multi-Texturing) 2511.5 MTexels/s

Currently stock HSF and cooling are used on GPU and SGRAM.








===============================================
Congrats. Very nice overclocking, for both CPU and FSB. Picture perfect !!!

Is there anything that may be a limiting factor now? What do you think any further improvement can be made, may be just a % or so?

What is your room temp and system ambient temp?

Any South Bridge cooling, the FSB may go a few MHz higher, if the Corsair 3200 LL let you do it.

How's is the PSU 12V line fluctuation?

BTW, what is reprogramming the zapped bios, what extra boost?

hitech... if you can tell me how or where to post some pics, i'll load them up... I'm not to keen on posting pics here since i don't have that privaledge, however, yes.... I customized my rig (secondary) with a heatsink from PCToys for the Southbridge, It's a nice blue aluminum, I got a fan for that heatsink from radioshack, (believe it or not) which still sells the HS for the 486's... my luck it fit perfectly over the blueHS from PCToys. I used the adhesive glue that came with the hs and its working perfectly so far. I glued the fan with silicon rubber sealer, which worked to my surprise.
My room temps are about 78degrees F... central air...(my house ofcourse) the 12v does fluctuate a bit more than I like, but I believe the culprit is that damn Turbine jet engine on my HS...
12v flux's around 11.86 to 11.92. loaded temps with folding@home is 45.50c (113F)

What zapped my bios is the 1st place? Case of the dumbass I guess. Looked at the abit site for updates, for which I saw that the 1.9bios was out for the NF7-S.... needless to say it was for the 1.2v of the board. That and after rebooting nothing came onto the screen. The emergency kit from Excaliber came in today,
(paid the 2day express) and followed the instructions. Was worried about pulling out a bios chip from my board while it was still on, but it worked. After which I was amazed that this mobo stood it's grounds at 220FSB which I couldn't get to on previous attemps. go figure

My system ambient is 29c
sorry for not including it.

just bumped up the fsb to 222mhz.... running folding@home without any problems... still at 40c idle, 45c load... coolage huh?

RedDawg41 said:

My system ambient is 29c
sorry for not including it.

Click to expand...

no wonder...try having ambience temps like me and see how well running 1.9v's is ...my ambience temp right now (1am) is 33c and my case temps are 35c...and this is with 3x delta 80mm screamers and 3x Tt2's and a slk-800a w/92mm Tornado...

Don't asked me about my a/c bill.... that's a kicker on my left nut...
Hell ain't this hot as it is in Texas, however them builders got the installation right on this part of the house, for which I thankthem! rigs are setup in the front where the dining room would of gone into ! ! !heheheeh (wife dislikes, but I pay the bills) !!

ran 2.500GHz @200FSB with v1.85 temps are 37c idle, 41c load running HOTCPUtester and folding@home. not to shabby. ambient temp inside case 27c.

Hmmm...

Maybe I should try to corrupt my BIOS with v1.9 so I can get a better overclock after the rebuild

BTW RedDawg41, congrats on your third star

Dude, I have no idea why this is happening after the corrupting the bios (self inflecting wound) but I don't know if it was that kit from Excaliberpc, or reprogramming the emptied bios chip which did it... but i wouldn't recommend this to anyone! DOH!

I was just joking of course

When I can upgrade to an SLK900U I'll be able to push more voltage at it. 1.55v/200/2100 ain't too shabby for a 30c ambient temp and a Vantec Areoflow.

The aeroflow worked out great for me right up to 2490MHz.... 42idle i think and around 48.5c loaded... v1.92vcore.... so i was pushing the limit on that HS... but now it's just a item of joy for me to look at on my desk.... whooo hooo ! !

got to get out more often.

I've got the Areoflow installed correctly and the CPU fully loaded. Right now I'm at my limits heatwise and rock stable. I can't increase vcore until I get better cooling. Durring cooler days I can go to 1.6v and O/C 200/2200. Even at 1.65v I didn't see enough stability to stay at 200/2300.

[email protected]
My socket temp reads 50C under load but the on-die temp sensor reads 58C. This is w/ an ambient temp of around 26C-27C.
Not exactly cool, but stable under prime95. Must say, I do like the Aeroflow.

believe me, If vantec had designed the aeroflow with a more efficient design technique (maybe all copper w/skieved fins) and keep the TMD with alittle more push of air (40-45cfm) then we'd be talking about a very competitive cooler... this monstrasity of the slk-900u with the tornado fan is just awful to the ears... but man it can cool that wacky cpu down.... v1.92 @40c idle, 45.5c load with folding@home... i'm @222FSB x11.5... 2.560GHz.... oh yeah this rocks, but I can't hear any more....

I did a calculation for the HSF w/ 0.3 C/W thermal resistance such as Vantec aeroflow. It loses about 200-300 MHz maximum overclocking compared to a SLK-800U on a Tbred B 1700+ DLT3C.

hitechjb1 said:

...
Here we can see how the overclock result of two different HSF's, since the one with higher thermal resistance limit the max power to the CPU for a given max die temperature. A difference of 212-300 MHz between a top end and a medium good HSF.
...

Click to expand...

Comparing different HSF's and impact of ambient temperature on max CPU frequency (HSF 0.3 C/W) (page 6)

That my certainly be true on paper, however, the areoflow itself maintained a constant degree of cooling up to its own limit, that's why, considering the rest of my components within my rig, the vantec was able to hold out at 2496+- 5mhz... with v1.95.... although, I never got to the 200FSB with the vantec HS, but that seemed to be another issue. I know this, I'm running 222FSB, @11.5multi... @2561MHz... which the difference between the vantec is only 69MHz.... so, it's on paper that this HS would appear to not give you enough cooling pwr, but with the right components/ambient teimps/xp1700 0319, it proved otherwise.