k-ohashi PentiumM755＠2640MHz 3DMark, AquaMark3

[sxs112]

M/B: AOpen i855GMEm-LFS / BIOS: R1.01
CPU: Pentium M 755 SL7EM/7416A480
CPU Cooling: MCX4000 + 2mmAG + DELTA FFB0812EHE
Chipset Cooling: 無改造
Memory: SanMax BH-5(339WF)B-V 512MB x1
VGA: GeForce6800Ultra
HDD: HITACHI HDS722540VLAT20
Power Supply: FSP400-60GN
Vcore: 1.476V
Vddr: 3.177V, Vio: 3.522V
OS: Windows XP SP1a + DirectX9.0a/9.0c
MEM: CL2-5-2-2, 400/266/133/267 MHz set(3:4)

edit: DDR-BOOSTER use、空冷約20℃

PentiumM755＠2640MHz=176MHzx15 ＆ 6800U＠498MHz/1.25GHz、
　3DMark2001SE　　　　30703
　3DMark03　　　　　　15527
　AquaMark3　　　　88441


3DM2001SE_30702


3DM03_15527


3DM05_6047


AquaMark3_88441


FF11_8962

source

[Kunaak]

it amazes me how little intel has done to follow this CPU up.
if I ran intel, this CPU would look like a gold mine to me.
I'd sell it as a gamers CPU, and sell them to alienware or something, and OEM.

[caater]

anyone knows what a64 clocks are needed to 30k with 6800?
i found these at ORB:
http://service.futuremark.com/compare?2k1=8059970
http://service.futuremark.com/compare?2k1=8101401

fx @ 3ghz and 3500+ @ 2.9ghz..

[Mathias-K]

Impressive results!!

But CPU-Z says the processor socket is "mPGA-479M". Well, according to Intel's own Processor Spec Finder, the only 2.0GHz stepping is the SL7EM, which k-ohashi is using here. The Spec Finder lists the SL7EM to be "478 pin PPGA".

Does this mean, that all 4 package types are compatible with the Aopen board? I'm talking about:

478 pin PPGA
479 pin H-PBGA
479 pin H-PBGA FC-BGA2
478 pin PPGA FC-PGA2

Now THAT would rock

[charlie]

real nice..........
yeah, if Intel had 1/2 a brain, they'd push the Dothan's as gaming cpu's....maybe even release a few new "flavors" like 2.2 and 2.4, to be followed up by a 200fsb/dual channel variety.

[eva2000]

fyi 2.1ghz dothan in japan was just released

[Shade00]

Absolutely incredible. It looks at least as good as the A64 in 3dmark, if not better.

[Kanavit]

dam dat good.

[tombman]

WAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH!!!

29sec @ 2.5ghz !!!!!!!

[macci]

2001SE would probably be at or above 32k with X800XT PE. The P4EE is starting to look like crap when compared to these CPUs

29sec @ 2.5ghz !!!!!!!

Yup thats just crazy

FX-53 at similar clocks hits 32sec [well similar is not a good term since I was useing 98MHz higher FSB, 39MHz higher RAM speed and DualChannel RAM in order to reach 32sec @2.5G (9x276.5, 1:1, 2-2-2-5-7-16)]

[cpulloverclock]

2001SE would probably be at or above 32k with X800XT PE. The P4EE is starting to look like crap when compared to these CPUs


Yup thats just crazy

FX-53 at similar clocks hits 32sec [well similar is not a good term since I was useing 98MHz higher FSB, 39MHz higher RAM speed and DualChannel RAM in order to reach 32sec @2.5G (9x276.5, 1:1, 2-2-2-5-7-16)]

I have ever seen 30s at 2.5 I believe
31s at 2.4Ghz

[tombman]

Yup thats just crazy

the only thing holding me back throw away my FX53 and 3.4 P4EE is the sad fact that there is no pci-e SLI mobo for P-M. Else P-M would be in my sys asap

[Mathias-K]

I have ever seen 30s at 2.5 I believe
31s at 2.4Ghz

Wow, if you can shave 2sec off af 32s FX @ 276 2-2-2 at the same clock you have some NASTY ram



These little beasties rock! I'd very much like some 21W cooking

[saaya]

very nice!

thx for posting sxs!

[jan]

yesterday i ordered the aopen board. i hope it'll be available soon.

cpu will be a 1.8 dothan es


this results are just amazing

[saaya]

still have that chip?

[drunkenmaster]

just a few things to mention, its a low fsb designed small pipeline core, extra bandwidth in so far as dual channel won't give similar to p4 style boost. Also we have no idea if the voltage he's pushing will kill the core in the next month, in which case that overclock might in effect be useless to people who want to keep them, also the 2Ghz dothan is very expensive.

HAve to say macci, 32 seems slow as my 3500+ can do 33secs at 2560Mhz or something like that, with ballistixs, the 30seconds CPULL seems a bit more on the mark for a ath fx.

Catter, those 3dmarks you posted are decent, fx is at 200Mhz stock fsb though which is killing the score, 2nd is a less cache, decent fsb but awful timings, 267Mhs cas 2.5,4,4,8 so still a little low, 30Mhz less on gpu core too. So both those rigs should bench quite a bit faster than this setup with right mem/fsb's.

Does look promising though. IT does make you wonder why intel aren't jumping straight too it. Several things i would suggest are, the P4 is cheaper/much higher yeild to make, and not 2mb cache so actually should be way cheaper. They are selling P4's by the bucket load even if they are slower in a lot of area's joe public doesn't know that and intel's millions and millions in advertising and being in bed with dell and many others are working well for them.

P4 might just be easier to work into a dual core setup soon and if amd have "dual core" in their available systems you sure as heck know intel wanna be able to say the same asap. Theres also the question they want to avoid of "whats wrong with the P4, why are you using your laptop cpu for desktop power computers?" Laptops in general are thought to be light, slower and portable, not known for power so switching will be a bad PR move.

HEy, if dudes chip is running in another 2 months then i'd consider buying one, but also only if the 1.6Ghz versions are clocking well too. I won't touch a AMD £300+ cpu and neither will i touch a intel one either.

My honest opinion would be that if intel could make this into a dual core cpu by next year they would, its slated for the year after and i doubt we'll see proper mainstream mobo support before then.

[drunkenmaster]

unrealneo, same manufacturing process, no, same manufacturing average size of transitor, yes. Its a completely different chip , voltage's a chip can take aren't down to size(its a factor) but the way the transistors are laid out and a whole bunch of other things. 0.13nm didn't mean that P4's could take the same voltages a Athlon 64 could, cos the cores themselves are too different. If theres anything you can say about dothan and prescott, its that they are very very different cores.

[drunkenmaster]

also, unlike a high wattage chip i wouldn't expect , better way to say it might be i wouldn't be surprised, if phase change didn't make nearly as much impact when using much lower heat outputs. I mean a MCX4000 + a delta can cool 100W+ of cpu very very easily, think what it does to a chip at half or less of that wattage?

Another thing would be , heat loads on phase change units could become an issue couldn't they? temps becoming much lower with same power but only 50% or less of the heatload? Obviously thats a very tweakable thing, but wondering myself(cos i don't know) would phase change units need changing for a much diff heatload?

[jjcom]

Yes but the P-M is more of a Pentium 3 than a Pentium 4...if only Intel had been smart....

[Hallowed]

Considering the sales of the P4, Intel was smart.

Whats faster, and what sells well, do not necessarily go hand in hand.

[jan]

still have that chip?

No. I'll get a new 1.8ES. This time in pga package -> Pins.

[saaya]

we have no idea if the voltage he's pushing will kill the core in the next month, in which case that overclock might in effect be useless to people who want to keep them, also the 2Ghz dothan is very expensive.

1.32v?

the P4 is cheaper/much higher yeild to make, and not 2mb cache so actually should be way cheaper.

$440 - Pentium 4 560 3.6GHz
$275 - Pentium 4 550 3.4GHz
$199 - Pentium 4 540 3.2GHz
$179 - Pentium 4 530 3.0GHz
$158 - Pentium 4 520 2.8GHz

$449 - Pentium M 755 2.0GHz
$313 - Pentium M 745 1.9GHz
$259 - Pentium M 735 1.7GHz
$217 - Pentium M 725 1.6GHz
$216 - Pentium M 715 1.5GHz

not that cheaper... im sure they would be a lot cheaper if intel would make more of them and sell them in retail!

dothan with 2mb L2 cache is still smaller than aprescott with 1mb L2 cache!

They are selling P4's by the bucket load even if they are slower in a lot of area's joe public doesn't know that and intel's millions and millions in advertising and being in bed with dell and many others are working well for them.

sounds like it :/

P4 might just be easier to work into a dual core setup soon and if amd have "dual core" in their available systems you sure as heck know intel wanna be able to say the same asap. Theres also the question they want to avoid of "whats wrong with the P4, why are you using your laptop cpu for desktop power computers?" Laptops in general are thought to be light, slower and portable, not known for power so switching will be a bad PR move.

dont think so... going dual core means you have half the bus bandwidth per core, and you have one architecture that doesnt need a lot of bandwidth and another one that needs it to perform nicely... wich one do you pick?

going dual core means double the heat, you have one architecture that runs cool (21W for 2ghz!!!!) and another one that runs hot (125W for 3.6ghz!!!!!!!) so wich one do you pick?

but i think your right, its all about marketing, wich of course, is really stupid.
you have a faster better cpu but dont sell it because it would mean some customers heads might explode after all the BS you told them in comercials before (mhz matters! centrino=not desktop compatible)

thats really stupid... if they wouldnt have such a high market share they couldnt allow themselves to do something silly like this.

HEy, if dudes chip is running in another 2 months then i'd consider buying one, but also only if the 1.6Ghz versions are clocking well too. I won't touch a AMD £300+ cpu and neither will i touch a intel one either.

My honest opinion would be that if intel could make this into a dual core cpu by next year they would, its slated for the year after and i doubt we'll see proper mainstream mobo support before then.

why wait 2 months? 1.32v dude... :P

unrealneo, same manufacturing process, no, same manufacturing average size of transitor, yes. Its a completely different chip , voltage's a chip can take aren't down to size(its a factor) but the way the transistors are laid out and a whole bunch of other things. 0.13nm didn't mean that P4's could take the same voltages a Athlon 64 could, cos the cores themselves are too different. If theres anything you can say about dothan and prescott, its that they are very very different cores.

pentium M = p6+ = tualatin+P4 bus +X
tuallies could take high vcores very well... i doubt they are more sensitive to vcore than prescotts...

Another thing would be , heat loads on phase change units could become an issue couldn't they? temps becoming much lower with same power but only 50% or less of the heatload? Obviously thats a very tweakable thing, but wondering myself(cos i don't know) would phase change units need changing for a much diff heatload?

you mean you think they run too cool for phase change?

Considering the sales of the P4, Intel was smart.

Whats faster, and what sells well, do not necessarily go hand in hand.

we dont know how the sales would have went with a P6+ core instead of netburst...

No. I'll get a new 1.8ES. This time in pga package -> Pins.

LOL!!!!

[Romaster]

hehe hollowed you got him there

last i checked the dothans costed around $400... that is quite costly

[macci]

HAve to say macci, 32 seems slow as my 3500+ can do 33secs at 2560Mhz or something like that, with ballistixs,

32sec at 2489MHz is slow if you get 33sec at 2560?? I think I got the 32sec already at 2430 too but since SPi 1M reports the result only in full seconds its a bit difficult to say how much difference there really is. At 2560 I can see an easy 31sec and 2sec difference between fx and a64 seems about right does it not?
also there can be almost 1sec extra difference between 'almost 31sec at 2488mhz' and 'just enough to hit 33sec at 2560'
I guess the FX could hit 31sec at lower clock if 'uber superpi tweaks' were used.
Hell in any case the Dothan beats the FX clock to clock as far as SPi 1M goes