XS HWBot Team

[SteveRo]

Ok - making some progress with the dominator 2kc7's
This is with MCH core at 1.48v and dramv at 1.78v.
cpu @ 4300, no oc on the vc, no Areca, x25-v as the boot and HDD target drive
Not done yet but doing much better!
Thanks for all the help guys!

http://service.futuremark.com/compare?pcm05=2349582

[Linuxfan]

Keep pushing SteveRo!
Nice work on that Q6600!

[G.Foyle]

Needs moar FSB!!!11one

As for DDR3 and P45 - LGA775 chipsets sometimes need high vDIMM for high FSB/memory clocks, regardless if the memory is capable of doing the same at lower volts. So new gen 1,65 V sticks are not a good choice usually. Unless you have something based on Elpida Hypers that can take 2,2 V daily and not degrade (GoodRAM PRO ).
2 GB sticks are also sub-optimal. I've had best results on P45 with 2x1 GB D9 Dominators (1000 7-6-5-18 in 3dm01).

[SteveRo]

Good morning Mr. Foyle, Yes, I suspected this ram was not optimal but its what I have and its probably better than the value ram I was using before.
I am currenty at FSB speed of 1720, what would you recommend I try and what MCH/dram voltages would you recommmend?
Appreciate all the help!

Mr Linuxfan, much thanks.

[G.Foyle]

As much as the CPU can do - high FSB will increase memory bandwidth dramatically. Try around 510 x 8,5 with 3:5 memory divider, if the CPU is capable of such high FSB.

[Hoss331]

The 3.33 (3:5) divider is borked, it doesnt run at its rated speed and is really only a 3.00 (2:3) divider. The only real option for higher fsb is the 3.20B (5:8).

[Don_Dan]

Ok - making some progress with the dominator 2kc7's
This is with MCH core at 1.48v and dramv at 1.78v.
cpu @ 4300, no oc on the vc, no Areca, x25-v as the boot and HDD target drive
Not done yet but doing much better!
Thanks for all the help guys!

Nice work!
Did you lose some FSB frequency, or can you still go up to 450MHz like in your previous result?
Any chance you can run 7-7-7-24 and tighten down the PL to 8 at those clocks? PL7 is probably too much to ask for from that poor little NB.

As much as the CPU can do - high FSB will increase memory bandwidth dramatically. Try around 510 x 8,5 with 3:5 memory divider, if the CPU is capable of such high FSB.

I doubt the CPU can do that kind of FSB speed, most of the other Celerons have even lower FSB walls, this one is quite good in fact.
It's worth a try of course, but I'm not very confident it's possible.

The 3.33 (3:5) divider is borked, it doesnt run at its rated speed and is really only a 3.00 (2:3) divider. The only real option for higher fsb is the 3.20B (5:8).

It's only bugged on the 400 strap, isn't it? However, running the 200 strap should be pretty hard on the board, so 5:8 is probably the best choice when using higher FSB speeds.

[SteveRo]

Thanks guys, you're the best!
Making progress finally! - http://service.futuremark.com/compare?pcm05=2350100

[Hoss331]

It's only bugged on the 400 strap, isn't it? However, running the 200 strap should be pretty hard on the board, so 5:8 is probably the best choice when using higher FSB speeds.

Yea supposedly its only the 400 strap 3:5 thats bugged but unfortunately they didnt include the 200 strap 3:5 on the ddr3 board.

[G.Foyle]

Thanks guys, your the best!
Making progress finally! - http://service.futuremark.com/compare?pcm05=2350100

Excellent now get that XP startup where it should be also, YGPM.

[G.Foyle]

Thankfully LGA775 days are over and I can forget all that arcane knowledge :P

[trans am]

UNcle steve kicking ass! Steve we need the almighty celeron 356 if we're gonna take this. I'm looking on ebay. I knew I shoulda never sold that rampage formula! I still have a p5k3 dlx. steve does your board have 2x16x? Looking back now I went about this contest the wrong way. I hastily went on newegg and found the only single core 775 chip 430 and with a multi of 9. that was retarded on my part. In retrospect ebay is the answ. screw the amd 140-145. that chip tops off at 5.6-6ghz if lucky and you would need the biostar ddr2 board Chew has to hit any decent HT at 400 or above. the ddr3 amd boards cant do that 400. he explained it to me when I benched with him a couple weeks ago. he's right. if you look at the amd dudes on ddr3 boards they can only hit those clocks with the Black cpus with the unlocked multis on ddr3. If you want to take this thing on ddr3 on single core then it has to be cedar mill and x48 or similar. Think about it. If I had a x48 board that has true x16x x16 slots I could use the lsi 9260 with the fastpath key and 3x 50gb vertex2 in raid0 and run that off the 2nd x16 slot. get a decent 356 cedarmill on ln2 and we would have this whole damn thing in no time. FACK! I wonder if our brother Paul (computurd) has any of his old 775 crap. if I remember correctly he had the WS and a bunch of other crap.

I know its been a while dudes but take a look at the single core king.

http://hwbot.org/hardware/processor/...tab=2drankings

steve you are one multi up from what I had with the 430 and at a measly 4.300 @ 1.9v on your cascade. wouldnt it be easier with x25 multi? you get the picture...
we went about this thing all wrong.

any xs dudes with old cedar mills get your asses in here!

THis pc mark stage is simple:
all you need to win this is some ln2,
a cedar mill 356 or similar,
some decent ddr3,
a board with x16 x 16 or a x16 and x8 slot,
a lsi 9260-4i/8i or steves ram drive
and a 5850 or higher,
some beer.

[Linuxfan]

THis pc mark stage is simple:
all you need to win this is some ln2,
a cedar mill 356 or similar,
some decent ddr3,
a board with x16 x 16 or a x16 and x8 slot,
a lsi 9260-4i/8i or steves ram drive
and a 5850 or higher,
some beer.

Thats not so simple TBH.... Tweaking for PCM05 is a must...

[G.Foyle]

You're forgetting that Cedar Mills are slow as hell clock for clock. I still think Sempron 140 can take the win and I'm gonna try to prove it

[chew*]

steve you are one multi up from what I had with the 430 and at a measly 4.300 @ 1.9v on your cascade. wouldnt it be easier with x25 multi?

Patience you must have young padawan.

This is chess not checkers.

[SteveRo]

... I still think Sempron 140 can take the win and I'm gonna try to prove it

I think you might be right.
Comparing several celeron vs sempron transparent windows scores, -
it looks like - at the same clocks - sempron is about twice as fast as the celerons.
All else equal, a sempron based system should win the pcm05 stage.
That said, am I the only one using an Areca/Acard setup for storage?

edit - the GGBT EP45T mobo drops to 8x/8x when both pcie's are occupied.
I think that is hurting me in my vga scores.

[SteveRo]

Guys - question - is H/W sharing amongst the team allowed for the HWBOT OC Challenge?
If there was a team member close to me that had a sempron setup, i could take my Areca/Acards to him?

[G.Foyle]

Technically speaking, PCMark05 is a 2D test, so you can't share a CPU, but you can share everything else, including storage setups.

[chew*]

Technically speaking, PCMark05 is a 2D test, so you can't share a CPU, but you can share everything else, including storage setups.

Simple terms it would be his score not yours.

Playing with ddr 3 board now, HT is looking promising

[G.Foyle]

If wish my C4F liked Semprons

[SteveRo]

It looks like (at single core speeds anyway) 8x/8x pcie is not impacting vga scores after all -

compare -

http://service.futuremark.com/compare?pcm05=2350100
- score 13806 with Areca 1880 in the 2nd pcie slot

to

http://service.futuremark.com/compare?pcm05=2350091
- score 10781 w/o Areca - all other settings kept constant

Also - looks like the areca/acard setup gets me something like 3k points

edit - over an x25-v that is (irams are probably much closer)

[SteveRo]

Simple terms it would be his score not yours.

Playing with ddr 3 board now, HT is looking promising

what proc are you using?

[G.Foyle]

It must be the general weakness of LGA775 then. If only Conroe Celerons did 600 fsb...

[trans am]

It looks like (at single core speeds anyway) 8x/8x pcie is not impacting vga scores after all -

compare -

http://service.futuremark.com/compare?pcm05=2350100
- score 13806 with Areca 1880 in the 2nd pcie slot

to

http://service.futuremark.com/compare?pcm05=2350091
- score 10781 w/o Areca - all other settings kept constant

Also - looks like the areca/acard setup gets me something like 3k points

edit - over an x25-v that is (irams are probably much closer)

Steve the x8 x8 x16 issue is good to know. Thanks for posting that. yeah the hit is not a big one at all. I ordered another 145 and a 2nd am3 board to try this on. You guys are right. I never thought about how efficient the architecture has improved since the time of 775. I ordered a board and another 145. good luck to you all if you wanna share a tip I'd love to learn more about pc2005. I never intend on winning these contests. I know I suck but I still wanna thrash with the heavy dudes.

[Don_Dan]

If wish my C4F liked Semprons

Nice score!

Congrats for taking down the 1x WR!

We're back in the lead in HWBot's OC Challenge October!



PS:

Thankfully LGA775 days are over and I can forget all that arcane knowledge :P

LGA775 will never die!