AMD Single-Channel World Record: 3516.66MHz - Conrad's Socket 754 Monster

[ozzimark]

These are a lot of firsts for me: first time with phase change, first time with DI, and first time with LN2. I'm just hoping that this CPU won't bug out at temperatures colder than -50C.

this should be exciting then!

hopefully you'll be able to push that little guy way past 3.5ghz and make the s754 record untouchable for a while

[Luising]

wow i want a 64

[conrad.maranan]

Well, I just got back from mWave to pick up another board. I'm hoping this will be my ticket to finally being able to run some 3D benches. I'd like to take another day off to play, but calling in two days in one week is just pushing it.

I should have some results to share by tomorrow.

[Vandi423]

oooooh! You actually picked it up? Hmmmm that might be useful.

So your other board is bad or something?

[conrad.maranan]

Yeah. I posted about it briefly somewhere in this thread. And then I dedicated a thread to focus on my situation over in the 3D section of the forum. In a nuttshell, I wasn't able to complete any 3D application - the computer would just shut down. After trying everything I could think of, I finally gave up and decided to start measuring voltages on my board. It turns out that I couldn't get a DIMM voltage reading from any of the spots that I usually measure voltage from.

So maybe my other 512MB module of VX isn't truly dead. Maybe there's hope. But I still think I'm going to purchase some new memory. Mushkin Redline is calling me.

[ozzimark]

Maybe there's hope

there's always hope

i take it you're gonna relax on trying to get higher screens for now, since you have the WR, and take the time to get the video card and ram working right?

[conrad.maranan]

Well, the bad news is still the whole 3DMark issue. I installed the new board last night, flashed it over to a BIOS that supports Revision E processors, and I still can't get 3DMark to run. I'm lost when it comes to this dilemma.

What are the chances of a Newark 3700+ being allergic to an X850 XTPE? This is the last thing I can imagine.

[Always]

Maybe it's a silly/noob question but this CPU of yours Conrad, is it actually a Turion ML-37?

[ozzimark]

What are the chances of a Newark 3700+ being allergic to an X850 XTPE? This is the last thing I can imagine.

that seems rather unlikely. it's possible that those two little chips actually did something.. but i'm not sure what. have you tried running hard drive benches and stuff?

always: nope, it's a mobile 3700+, the ML-37 is clocked at 2.0ghz stock, and uses the lancaster core, which was shown to be a somewhat lame overclocker

[conrad.maranan]

Perhaps those two broken chips are affecting something. It's just strange that I can run 3DMark with the crappy card, though. Either way, thanks for accepting my request to test the card if I have to go that route.

[ozzimark]

i'm honored conrad, and i'm also very eager to get to the bottom of this

[conrad.maranan]

I was listening to some old LL and something happened...

Don't you dare stare
You better move
Don't ever compare
Me to the rest that are all gettin' sliced and diced
Competition's payin' the price

I'm gonna knock you out!

Mama said, "Knock you out!"

I'm gonna knock you out!

Mama said, "Knock you out!"

Breakdown!

CPU-Z Validated at 3516.66MHz


The good news:
- My other 512MB OCZ VX module miraculously came back to life
- I finally broke the 3.5GHz barrier - this time with 1GB of RAM at 251.2MHz 2-2-2-5-1T
- I tested my card with my sister-in-law's Clawhammer and it runs 3D apps fine

The bad news:
- I still can't run any 3D apps with my card using the Newark 3700+

[Jupiler]

Awesome clocks, conrad.

To bad the 3dmark problems still persist. Would love to see some 3d action with that setup.

[piotres]

I was listening to some old LL and something happened...

CPU-Z Validated at 3516.66MHz


The good news:
- My other 512MB OCZ VX module miraculously came back to life
- I finally broke the 3.5GHz barrier - this time with 1GB of RAM at 251.2MHz 2-2-2-5-1T
- I tested my card with my sister-in-law's Clawhammer and it runs 3D apps fine

The bad news:
- I still can't run any 3D apps with my card using the Newark 3700+

Wow awesome OC ...are you still using chilly1 single-stage ??

[ozzimark]

- I finally broke the 3.5GHz barrier - this time with 1GB of RAM at 251.2MHz 2-2-2-5-1T

YES!!!

i knew you could do it!


as a side note.. ouch, 1.85v
i guess it can take it if you keep the temps down

[[XC] moddolicous]

See, I knew u could do it. Never lost faith. Great OC BTW. Can u run SPI?

[Vandi423]

Great Job Conrad!

I would bet that your problems lie with those two chips. I am surprized that it even got past 3.0GHz. Too bad I don't have any extra Rev.E processors or else I'd send you one.

[conrad.maranan]

BIOS is set to 1.475 + 126% for that screenshot. I'm still on the same phase change unit so there's probably more to be gained with cooler temps. I'll run SuperPI later on and see if anything has changed from the runs in my first posts.

It's pretty sad having one of the best video cards installed in my system, yet it's only capable of doing the same thing that a used $5.00 PCI card can do.

Vandi423 - When do you expect to have some again? PM me, bro.

[ozzimark]

that's sooo much voltage for a 90nm core. how long you think it'll last with that?

i'm liking that little bit in your sig now

[conrad.maranan]

Thanks, ozzimark. As far as the voltage is concerned, I'm not afraid. I'm going to have to purchase another CPU anyway. Hopefully, the next chip will be just as good of an exhibitionist as this one. If not, this one can be my "suicide" chip and the other one can be my "3DMark/Multipurpose" chip.

[EMC2]

Darn bro... what next, 3600? Very nicely done

A suggestion regarding your 3D problem bro... try lowering the HT link multiplier (just to see, even set it to 1x)... your broken parts maybe affecting your HT link

[i found nemo]

ok i don't visit these parts of the forum very often and i just seen this....conrad why din't you forward me the link??? anyways man good job as far as 3d mark freezes, why don't you get a 2600+ and test that with the same setup, just to see what's goin' on if indeed it's the chip, play around with diff rams in diff slots.....btw ur cpu pwns mine

[conrad.maranan]

Daniel - I've tried the full range of options on both boards - everything you can possibly imagine. No joy. Do you think that the two chips I broke have something to do with the way voltages are being regulated? If a crappy AGP card in the same system can run 3D, is it safe to assume that a card with greater power requirements is being affected by the CPU's inability to regulate voltages? I obviously know nothing on the subject of electrical hooplah.


Jason - No need for a 2600+, bro. 3D works with my sister-in-law's Clawhammer on the same board. 3D also works with a crappy video card on my system. I've done everything you could think of - minus the substitution of my current CPU with the exact same one, but without the broken chips. And of course my CPU pwns yours. But if you're having fun with what you have, that's all that truly matters.

By the way, my new goal is 3650MHz with DI and 3800MHz with LN2. I'm such a big dreamer.

[EMC2]

Daniel - I've tried the full range of options on both boards - everything you can possibly imagine. No joy. Do you think that the two chips I broke have something to do with the way voltages are being regulated? If a crappy AGP card in the same system can run 3D, is it safe to assume that a card with greater power requirements is being affected by the CPU's inability to regulate voltages? I obviously know nothing on the subject of electrical hooplah.

Bummer No, has nothing to do with voltages... CPU's have nothing to do with regulating the power The parts broken on top are one of 2 things (picture wasn't clear enough to tell which)... either termination packs or bypass caps. Either one could affect the HT interface if they were hooked into that part of the circuitry of the CPU. That's why I suggested a lower HT multi - to get the HT link frequency way down to eliminate it as a possibility. Next time you have the evap off the CPU, if you can get me a really clear close up of the CPU (where I can read markings on the bits), I may be able to apply some Java and noodles to it It's obvious from your outstanding results that they aren't hooked into the memory interface.

I'll go back too and look again closely at the symptoms and see if anything pops out.

Peace

[edit] looked thru the thread... not enough description of the problem... define "system shuts down" - you mean BSOD, reboot on it's own, PS shuts off, ???? ... also, can you individually run all the phases one at a time without the problem or is it a certain phase it fails when running 3DMark? Are there any single 3DMark phases you can run in a loop for 30 mins?

[Supertim0r]

It sound like bad memory sticks
Like you said too, it could be the broken chips but ...
Do you have some other memory ?
Too bad you guys are in the US, i've got 28gig of Value VX @ work