Overclocking Opteron 175

[Biscuit]

Had this chip quite a while now and never thought to push it.

Currently running with 333mhz ram 10x250fsb 1.425Vcore

Orthos running CPU + ram stress for 10 mins was stable

Temps
Idle -
Speedfan = 38
Core temp =
core 0 - 36
core 1 - 38

speedfan = 56-58
Coretemp =
core 0 - 50/51
core 1 - 55-57

Seems a little high to me although i have just redone the heatpaste.

Previously i had only put a really small amount on however this time i thought i would put a bit more on as when i took it off it didnt look as though it had seated with the HSF verywell. This look ok to other people or should i take off the heatsink again and scrape a little off?

Any other advice people have?

[s_kinton]

An amount the size of a grain of rice is said to be correct when applying thermal paste.

The temps do seem a bit on the high side under load, but nothing to worry about. Just a bit above the ideal. Idle temps look good. What hsf are you using?

- Just increase the HT (what we used to think of as FSB on older amd chips) in increments and find the limit of stability using orthos.
- Make sure your ram is on a divider to rule that out.
- Only increase the vcore when you're sure you've got the max stable clock speed.
- Watch the temps always. Dont really wanna go above 60-65c
- Once you've found the max clock speed at an acceptable running temp, increase the ram speed to find your max overall speed.

[Biscuit]

i assumed that by setting the ram at 333mhz & the multiplier at 10x i had set the divider?
I have a themaltake bigtyphoon and im using AS5. I do have some issues with temperature though and i think they are mainly down to my 8800GTS and a lack in inwards airflow.

By overclocking the fsb frequency does that not automatically overclock the ram?

[s_kinton]

i assumed that by setting the ram at 333mhz & the multiplier at 10x i had set the divider?

By overclocking the fsb frequency does that not automatically overclock the ram?

Yes, you did set the divider.

By overclocking the HT (not FSB, FSB doesnt exist anymore) the ram will increase proportionally. So right now your ram should be running 383mhz (333+50). Check using CPU-Z.

Also, dont forget to reduce the HT multiplier to x4 up to 250 and down to x3 for anything above 250.

I would suggest you google for some a64 overclocking guides and follow one through. Theres plenty out there.

[NightshadowUK]

Orthos is FULL load, you'll not get that kind of strain even during heavy use. If you can keep Orthos temperatures under 65/60ºC you're perfectly safe.

[scrandman]

Yes, you did set the divider.

By overclocking the HT (not FSB, FSB doesnt exist anymore) the ram will increase proportionally. So right now your ram should be running 383mhz (333+50). Check using CPU-Z.

Also, dont forget to reduce the HT multiplier to x4 up to 250 and down to x3 for anything above 250.

I would suggest you google for some a64 overclocking guides and follow one through. Theres plenty out there.

You're partially correct. With A64 the RAM speed works on an overall divider from the CPU speed.

For the example in question with the CPU on 250 x 10 and the RAM on the DDR333 setting, the RAM actually runs at 208.33 (2500/12). Another example would be the CPU on 250 x 11 with the DDR333 setting, the RAM would actually run at 2500/14 = 196.43. As you can see, the actual divider applied is dependant upon the CPU multiplier.

Rather than remembering these dividers or trying to figure them out, there is a great little tool you can download called A64 Memfreq which allows you to plug in values and it automatically works out the resulting memory speed for you.

[notsobig]

Agree with Nightshadow, btw do you really need +0.75v vcore to achieve 2.5G? To me it seems bit too much to be honest, as I know many people running their opty @ 2.5-2.6 at stock voltage.

[CooperXS]

S&M FPU test is much harder on K8 than Orthos. Try it - you'll see even higher temps

Bisscuit

What HSF you are using?

[Biscuit]

im using a thermaltake big typhoon. Check my system for full list.

Tried running at stock volts but it wouldnt even boot up into bios, my psu maybe not up to it but when i cranked it upto 1.425 it runs fine. Did 4 hour orthos test and no issues.

BTW notsobig its only 0.25 over stock. stock vcore is 1.39 to 1.4 on x2 939s

CooperXS - so your saying if i try this test i may hit dangerous temps if im getting 60sih with orthos? maybe worth resorting back to my 400 memory 8x multiplier and 220 fsb which is what i was on b4 to give 2.4Ghz?

Currently at 333 mem div 260x10 = 2.6Ghz

stable but temps are 57-62 under orthos load

[CooperXS]

I'm saying this test is the best for testing overheating -IF- you will use your system 24/7 under full load. I haven't seen any application heating up CPU so mad as S&M does and I like to know that I still have some room in case of high ambient.

TTBT is good HSF, but really needs to have lot of intake from the side of the case and nice airflow just between the fins and motherboard to remove all hot air through the back of the case. Also this HSF comes with horrible base finish and requires some lapping to get better results. If you will do it try to make sure to keep it flat, since adding more thermal grease in those gaps will result in even worse contact. Also some people make different modifications to the HSF for better performance. You may find some on xtremesystems.
And remember that maximal operational temperature for this Opty is 65C

[NightshadowUK]

Stock vcore on an Opteron is 1.35v man.

[Biscuit]

Stock vcore on an Opteron is 1.35v man.

Mine has always run at 1.39 I might try going back to stock and setting volts at 1.35 see what happens

[dmuz]

I run my 170 at 1.35V/2.5G. I can't drop it to 1.325 tho

[Biscuit]

Got everything to seem stable at 2.65 1.375v but when i load a game it lags like mad. I think maybe my mobo is a big dodgy but, more likely, my psu isnt up to the job. Back with everything normal except fsb is set to 220. Aaahhh wel 2.4 will do for me anyway