Dragon meets Shrimps ... on ice!

[massman]

A few hours ago I dropped off Blind_ripper back at his home, so now I have some time to share Madshrimps' first Phenom II session. Long story short: we only recieved our Ph-II a couple of days ago, which didn't provide us that much time to prepare our setup. I posted a couple of results in the AMD section (4G 1M/32M), but that's about all the preperations we did.

Without much further ado, the results:


-1 core: http://valid.canardpc.com/show_oc.php?id=500625







I certainly don't have the best cpu around: 4 cores only uptill 5.5G maximum and adding vcore doesn't really help the cpu scale. Though, I had an 11.218s 1M run (almost had you there Sampsa!!) with a very nice efficiency (more on that topic later). I'll add some pictures tomorrow + a link to the article on www.madshrimps.be.

First: sleep!!

[blind_ripper]

whas a fun day of benching again with u !

[[XC] gomeler]

What week was the CPU? Retail or press sample? Nice clocks there It didn't scale with any additional voltage or is CPU-Z reading low?

[TiTON]

Great run.. any info on the batch?

Ton

[blind_ripper]

tomorrow guys massman is allready asleep

[Hyperhorn]

Great efficiency, well done!
No chance to get tighter timings @ 560 MHz+ because of the IMC? (tRAS that high? )

[blind_ripper]

the NB did not like more then 2.14vdimm, so no go on really going hard on the memory .

[Expat GriZ]

Nice Job GUYS!!! Go BELGIUM!!!

[V2-V3]

Very nice!

Looks to be a good chip, have you tried lower voltages?
maybe 1.8-1.825?

[Sampsa]

Nice results, it's 3 AM here, more tests tomorrow with SF3D and macci

[(TAT")]

Nice 1M result

btw, the fact that your NB doesn't like high vdimm may give us some hint on why our NB frequency could not reach above 4GHzcos we have not tried anything below 2.4V vdimm...

get some good memories and you will beat Sampsa's 11.234s

[massman]

What week was the CPU? Retail or press sample? Nice clocks there It didn't scale with any additional voltage or is CPU-Z reading low?

It was sent by AMD, but it's definitly not a press sample or a cherry picked one. The stepping is without any doubt retail.

CPU-Z is reading correctly. The sweetspot for this cpu was around 1.808-1.856-1.888v higher didn't gain us anything but instability. Note that the 5.6G 3D run was only one core, though incredibly stable.

Great run.. any info on the batch?

Ton

0849 BPEW 0223, not a bad one (1.488v needed for 4G 32M stable), but definitly not the best (only 2.4G NB on air).

Great efficiency, well done!
No chance to get tighter timings @ 560 MHz+ because of the IMC? (tRAS that high? )

The efficiency was easy to obtain, actually, anyone can do it. I figured out why my first couple of runs had an efficiency of 70k+ and I only had sub69k runs. I did one at 4G/4G and hit like 68151 efficency

I did indeed not tweak the memory, something for the near future

Looks to be a good chip, have you tried lower voltages?
maybe 1.8-1.825?

Yeap, tried, but nothing special in terms of gaining speed. The one core is pretty good, but not all four. Like I said, 5.4G is about the maximum I'll be able to run 3DM06.

Nice results, it's 3 AM here, more tests tomorrow with SF3D and macci

Hehe, can't wait to see what that'll give .

Nice 1M result

btw, the fact that your NB doesn't like high vdimm may give us some hint on why our NB frequency could not reach above 4GHzcos we have not tried anything below 2.4V vdimm...

get some good memories and you will beat Sampsa's 11.234s

My memories weren't that bad and I actually did beat Sampsa in some of my runs (11.234 and 11.218). The issue was that saving the score was rather impossible due to the memory being too cold, I reckon (or the board).

Oh, 2.4v might be too much. When I passed 2.140v (bios) I noticed a quick decrease in stability, which was indeed caused by the NB. Try to keep below 2.2v to catch 4G NB. I'll try whether 1.9v memory helps to pump up the NB clocks .

[(TAT")]

Oh, 2.4v might be too much. When I passed 2.140v (bios) I noticed a quick decrease in stability, which was indeed caused by the NB. Try to keep below 2.2v to catch 4G NB. I'll try whether 1.9v memory helps to pump up the NB clocks .

Thanks for your info, we'll try that in our next benching session

[Foamy]

Amd looks sexy these days!
Great effort from both of you guys.

[dread77]

Great results Massman!!! Your low NB frequency on air could be due to your motherboard. I have the exact same batch with you and I can do an easy 3000+ NB Frequency on air on an Asus M4A79 Deluxe board

[Lightman]

Nice set of scores Massman and Blind Ripper

I've noticed same behavior regarding NB/memVid.
Also my Core 1 (from 0,1,2,3) is better than the others but the difference is not that big on AIR, around 100MHz.

[massman]

Also my Core 1 (from 0,1,2,3) is better than the others but the difference is not that big on AIR, around 100MHz.

Seriously? It's the same here

I noticed quite a gap between Core 0,2,3 and Core 1 regarding stability. C1 did the 6.1G, C0 didn't do 6G and C3 was falling apart at 5.85G or so

[massman]

Great results Massman!!! Your low NB frequency on air could be due to your motherboard. I have the exact same batch with you and I can do an easy 3000+ NB Frequency on air on an Asus M4A79 Deluxe board

Hm, good to know. I have an M3A79 here as well, I might give it a go :-)

[Aerou]

This is incredible. The results are very professional and congrats, but just imagine, if AMD had such chips earlier - the stakes would be different.

But get that memory going more! It may not be such a factor as much as it leaves a weak impression 5-7-7, you know

Are you going to unleash the Dragon? ( I mean some 3Dm action soon?)

[subaruwrc]

Nice clocks, keep pushing

[massman]

This is incredible. The results are very professional and congrats, but just imagine, if AMD had such chips earlier - the stakes would be different.

But get that memory going more! It may not be such a factor as much as it leaves a weak impression 5-7-7, you know

Are you going to unleash the Dragon? ( I mean some 3Dm action soon?)

The 5-7-7 is due to the 3:8 memory divider, actually. But, yes, I need to spend some time tweaking those ...

I'm actually considering switching to my Core i7 rig for 3D action as it's, obviously, still faster in most cases. Petri and Sampsa did a fantastic job in 3DM05, but we need to remember that 6.3G stable was on Liquid Helium, not Nitrogen. We'll see how things go

[dduckquack]

way to go massman! amd looks like its fun to clock

[Bullet92]

Very nice job guys. Time to tweak that memory more and crash WR with some nice sub11s score
Looking forward to new (awesome) results.

[V2-V3]

with 1.7v core I was able to hold 5.3Ghz in 3Dmark 06 and Vantage, Ill pull up the batch and NB voltage l8r 4 ya

[Lightman]

Seriously? It's the same here

I noticed quite a gap between Core 0,2,3 and Core 1 regarding stability. C1 did the 6.1G, C0 didn't do 6G and C3 was falling apart at 5.85G or so

Maybe we can do some widespread survey and find out if this is a common behavior??

Are the Cores 0 and 1 closer to L3 Cache or further?
I wonder if K10 core w/o L3 cache will show similar OC disparity between cores. Time will tell.

PS. When efficiency article is coming up?