939 Winchester 3000+ 0441SPMW

[trans am]

"It could be strained vinegar or hyper osmosis purified rat piss for all we know. That's just my ."

ROFLMAO!



ADA3000DIK4BI
CBBGD 0439VPCW
1095967J40197[/QUOTE]

[qk4722]

My new cpu 939 3000+ 0441SPMW x4 :banana4:

[eva2000]

yuummmy.. nice spend

[trans am]

Nice, $1400 in CPU's Where were these from?

[boblemagnifique]

qk4722 : 3000+/3200+ CBBFD 0441 SPMW ?

[OC Detective]

It could also be an exceptionally good chip from a truly perfect piece of silicon crystal - purified rat piss only helps you until a certain point ...

I'm going to get a new 3200+ Friday or Saturday (the shipment is due to arrive at the shop on the 18th), that ought to be from a quite recent week, as K8 fabrication is done in Dresden (not too far from here) and there are 2 of the 12 german AMD distributors within spitting distance of me. Fingers are crossed, hoping for week 44 or 45.

Dont count on it - remember after they are fabbed they are sent to Malaysia for assembly and test then off to Singapore for final test before heading to the various distribution centres around the world. Typically Japan sees them first then US then Europe so you could just see a similar assembly dated cpu to those that people are getting in the US. Then again with the problems AMD have had rolling out new software at their distribution hubs who knows what you might get!

[johnnyliu3377]

very nice! 372Mhz qk4722

good job :thumbsup: , I also must try to it!!

[conrad.maranan]

Hmmm. I don't understand how running at 372MHz with a multiplier of 6 is impressive, especially with the RAM at 186MHz and 2.5-4-4 timings.

[Geforce4ti4200]

its impressive for the motherboard's super high HTT. I am really hoping my mobo will hit 300htt

[johnnyliu3377]

Hmmm. I don't understand how running at 372MHz with a multiplier of 6 is impressive, especially with the RAM at 186MHz and 2.5-4-4 timings.

I think ....This mainly is tests M/B(K8N neo2) the high-frequency, the memory is unable to achieve DDR744(1:1) in at present.

[conrad.maranan]

Now I understand. You were testing the highest HTT you can reach.

[qk4722]

qk4722 : 3000+/3200+ CBBFD 0441 SPMW ?

HTT316x9 ram 1:1 2.5-4-3-5 2.85v PI 29s :banana4:


ADA3000DIK4BI
CBBFD 0441SPMW
1077009K400XX

[krampak]

My friend have just arrived ^_^



Lets see how a week 42 performs :banana4:

[Geforce4ti4200]

On air cooling? 2.9GHz would be nice and with it being unlocked you can max out your ram first than keep rising the multipliers. 240x12=2.88GHz if your ram and cpu can do this

[funkflix]

240x12

... U mean 10x280 MHz!

[krampak]

Well, i've pluged it. I took a screenshot at 2'9Ghz 1'53v stock cooler, but this cpu is pretty hot... I get more than 60ºC at 1'6v full load :S (with the XP-90 too).
By now it's running at 2'75Ghz 1'55v.

[krampak]

Well.. i've opened my room's window.. it was a temp problem, just passed from 57º full to 49º full @ 1'50v, It has done the 8Mb superpi at 2'81Ghz 1'50v, not bad at all for a stock cooler
It does 3Ghz, Vapochill needed... it's getting sooooo hot! (an unstable).

[Hallowed]

I hate to put a crimp in the GF4 bash session, but its fairly well known that Winchesters are ~5% faster than Newcastle 939's per clock due to faster L2 cache.

Well-reputed websites have confirmed this, as well as personal testing from multiple parties.

Now the week 45 theory... :roll:

Anyway, excellent OC to the original poster.

[JNav89GT]

Winchesters are 5% faster than 754 Newcastles (at a push) they most certainly are not faster than 939 130nms! You need to get your head out of the sand on that one. Even the webpages you used as proof say otherwise, show me one site where it shows the overall performance is 5% better - I think I will have a long wait...........

they are faster, 5% I dunno. But faster yes. I reviewed against 0.13u Newcastle 939 and results were very very good.
http://www.madshrimps.be/?action=getarticle&articID=230

[OC Detective]

I hate to put a crimp in the GF4 bash session, but its fairly well known that Winchesters are ~5% faster than Newcastle 939's per clock due to faster L2 cache.

Well-reputed websites have confirmed this, as well as personal testing from multiple parties.

Now the week 45 theory... :roll:

Anyway, excellent OC to the original poster.

I have yet to see one site show a 5% overall improvement on a 939 90nm v 939 130nm if you refer to the Anandtech site go compare the 3500+ results overall (which are heavily gaming slanted) it comes out around 2%. There were no significant architectural improvements made to the 90nm over the 130nm - AMD said that themselves!
http://www.xbitlabs.com/news/cpu/dis...011124532.html
Just because it was bandied around before they were released on this site about the 5% does not make it true!
JNav when I made that comment I meant they are certainly not 5% faster than the 130nm. Weeks ago I went on record on other sites saying the upside is 1 - 2%.

[JNav89GT]

well until you have personally run each side by side, I would caution you to believe press releases over actually reported results. Please reference my article. Anandtech saw similar trends. The only review I saw where results were identical used an Asus A8v motherboard. At least on NF3 for me the results were impressive.

I will say that 5% is too optimistic. However, when gauging performance, the FX53 is barely 5% faster than a 3500+ when stratifying for clock speeds. 5.5% faster in 3dmark2001se per my calculations on my benchmarks w/my 3200+ review. Teh 3200+ was 3.5% faster than the 3500+ clock for clock. So it's closer to the FX53 than to the older 0.13u cpu. I don't believe they didn't tweak the core some. I bet they are just not letting on b/c there are more features to come that have not been implimented as of yet, as in SSE3 etc... Then they will announce and get people to want that new cpu.

[OC Detective]

Jnav I have just read your review and it completely bears out what I stated! When both are at 2.5Ghz the upside is between 1- 2% average overall (though some of your findings were strange - especially in UT!).
Why would AMD say there were no significant improvements if there were - wouldnt that be counter productive?

[JNav89GT]

yes but if you look and the number, the 3200+ bridges gap between the 0.13u and the FX series. Of course it can't catch the FX but for a few benches, but there is performance to be had. These were not flawed scores. I reran benches when I first got chip to confirm findings. I agree the % isn't always that impressive, but for a dye shrink there MUST be some improvements to account for these improvements. Like I say it splits difference between the 0.13u and the FX.
Why would AMD say no improvements? Beats me, other than they are waiting for new stepping where SSE3 will be ready and maybe even more core advancements so it will appear larger improvement in comparison to the 0.13u cpus. That's my theory, but take it as such, just a theory.
the UT numbers are wacked, esp with the 754 beating 939, but that I reran too. Not sure what to say other than in the comments after benches there.

p.s. I have bench screenies if you don't believe me

[OC Detective]

Dont worry I believe your scores like I said they average overall 1-2%. Actually Petr listed the changes before and they are not significant.

[Hallowed]

It depends if the certain benchmark is heavily dependent on cache speed.

Because thats the difference, _the only difference_ between the two, performance wise.

If the benchmark is reliant on other factors, or to lesser degrees, then yes the performance increase will not be as high, or even not present.

For example, using the Anandtech bench, the Winchester is roughly 4.5% faster in Quake 3, an app which more or less runs from cache. DOOM3 on the other hand only benefits 2.5%. 2001 also shows about 5% over Newcastles.

Nevertheless, heres why: