first venice and now this s7? good job!
Printable View
first venice and now this s7? good job!
Exactly what I was thinking, welcome back s7e9h3n!Quote:
Originally Posted by St Devious
Have you been in the future? :rofl:
BTW I love the Task Manager screenshot with 24threads!
As someone said, where in Europe are you?
Quote:
Originally Posted by justapost
Whoops--almost forgot about this. Here's a run at default for me. I think I must be doing something wrong. Not sure how strongly effected by ECC this is, either. I run with ECC set to maximum protection. Come to think of it, this may be with underclocked memory. Let me check...Quote:
Originally Posted by Mechromancer
http://www.pcrpg.org/pics/computer/opty2427/stream.png
Default: 12x2200, DDR2-800, ECC Off...not a whole lot better
http://www.pcrpg.org/pics/computer/opty2427/stream2.png
Ohh thanks, those results look like single socket vs. dual channel. :)Quote:
Originally Posted by Particle
Do you really have all four channels populated and ran with 12 threads? Beside that with six cores sharing two channels this benchmark is heavi imc speed limited. My 955be and also the AII 240 and 250 get best results if the nb runs one notch below the cpu. I hit a limit once the nb speed reaches 2xmem speed, so the optimum with ddr3 1333 is 2.8GHz/2.6GHz cpu/nb. If it's the same with DDR2 you should not be limited by the stock nb speed however since it's already above 1.6GHz.
Beside that I think only this ES has unlocked multis and voltages and the retail chips will be locked so using k10stats or similar to modify these settings on the chips msr register will not work and oc capability will be limited by the max possible ref HT the boards will offer.
@s7: I'm curious, what msr register did you modify? Was it the controll register or the p-state-0 register?
FASN8ing :up: (times 3).
I really hope some of these new G34/C32 boards will be made with "XS
people" in mind.
Definitely sounds like there is a 2P market here.
Many thanks for sharing S7. :clap:
- Dual Opteron 2427s (Istanbul) @ 2.2GHz CPU/2.2GHz NBQuote:
Originally Posted by justapost
- Each CPU has two 2GB DDR2-800 DIMMs installed in dual channel configuration
The numbers are just lower than I'd expect.
can you run 3dmark 06? :)
Particle, Maybe it has something to do with the chipset difference?
Check your numa settings in the bios you should get around 20MB/s in that benchmark, just like the other dual socket shanghai/istanbul setups.Quote:
Originally Posted by Particle
So someone finally noticed...
http://www.fudzilla.com/content/view/15552/1/
Hehe This thread started so long ago I'd completely forgotten to watch Fudo. I thought they'd pick it up quicker.Quote:
Originally Posted by Smartidiot89
Didn't they get it wrong though? this is 2x CPUs right? And all of the benches were run with 2 CPUs right?
Looks like the CPUs were definitely throttling:
http://img9.imageshack.us/img9/7927/wprime28.jpg
This run is over 1 sec faster than @ 3ghz.
Don't know how much I'll be playing with the setup today as I'm dead tired from my trip.
:clap: Great. Waiting for more.
Do you have better cooling on them now?Quote:
Originally Posted by s7e9h3n
Could you please test with WPrime v1.55, then it's easier to compare to hwbot. :)
can u try Cinebench run? And winrar benchmark?Quote:
Originally Posted by s7e9h3n
wprime is better with the same HT: NB (2000:2000, 2400:2400 etc). Example with 2000MHz HT and 2600MHz at NB is run whorse. But i think, u know it:)
WPrime 1.55 seems to have issues detecting the proper number of cores.....Quote:
Originally Posted by Don_Dan
http://img170.imageshack.us/img170/8762/wprime155.jpg
.....so does the Winrar bench:Quote:
Originally Posted by FlanK3r
http://img294.imageshack.us/img294/5459/winrart.jpg
Cinebench has the same problem as well...don't have a screenshot at them moment, but I'll post one later....
BTW, same stock coolers...just less Vcore ;)Quote:
Originally Posted by Don_Dan
You're way too ahead of the pack with your gear, software can't cope with you :DQuote:
Originally Posted by s7e9h3n
That 1 MB is inaccessible so no point seeing it as 2x 6 MB L3.Quote:
Originally Posted by Nedjo
have your tried setting it Manually on both ?Quote:
Originally Posted by s7e9h3n
Lol...no...I guess I'll try that now...thought these programs would be a bit "smarter than that :p:Quote:
Originally Posted by demonkevy666
I think he can't adjust it.Quote:
Originally Posted by FlanK3r
This is not an overclockers board......if he could adjust he wouldn't be writing scripts to OC all cores he would just flip multis in bios ;)
S7 advanced tab on wprime 1.55, set to 24 cores :)
Yep...frequencies aren't adjustable. Only multi:Quote:
Originally Posted by chew*
1.55 seems a lot slower:
http://img223.imageshack.us/img223/8140/wprime155a.jpg
Any programs out there that adjust memory timings?
I really wish codered still developed A64tweaker. Remember how vital that used to be for BH5 tweaking back in the day? I don't think anything aside from AOD allows you access and adjust to all the memory timings. And especially not for the board you're on.Quote:
Originally Posted by s7e9h3n
Try installing Amd Over Drive S7, no clue if it will work though, it might not detect AMD cpu :ROTF: if anything can fool with ram in the fly its that......might be able to raise HT bus, etc bump NB and HTT ;)
BTW if this is sleeping I want to see you Work.
lol! For real! S7 is pumping out benchies with one eye open it seems.Quote:
Originally Posted by chew*
Well, that was a complete failure...lol:Quote:
Originally Posted by chew*
http://img33.imageshack.us/img33/623/aodo.jpg
As always baredit/crystalcpu and the bkdg. :DQuote:
Originally Posted by s7e9h3n
I think he is as excited about it as we are and missed it a lot when he was away.Quote:
Originally Posted by chew*
Hey s7, this memory tweaker worked quite well on Phenom I's...Quote:
Originally Posted by s7e9h3n
Might be worth as shot... ;)
http://www.xtremesystems.org/forums/...ad.php?t=92190
I was going to say AOD for tweaking memory, but you already find out it's not working. Still it would be worthy giving a shot with AOD 2.1.4. That was the version which worked without knowing the CPU and while newer releases were crashing on my PC this one was happy bunny! The only drawback was lack of info in some tabs ...
EDIT: Also AOD feels best on XP or Vista OS ...
Have you tested CPU-Tweaker by FELIX?Quote:
Originally Posted by s7e9h3n
That setup has four memory controller, all those apps only support one controller.
Chapter 3.5 here shows the pci register specs.
http://support.amd.com/de/Processor_TechDocs/31116.pdf
In BarEdit their should be four pci devices for the different memory controllers. I tested it on an agena single socket system last year and it worked.
It's cooler today :D
http://img41.imageshack.us/img41/7678/wprime155b.jpg
I'm getting a bit tired of WPrime. I'm going to grab a 4890 and throw it in the system. Probably won't have results till the weekend, but I'll try to get some 3D runs in.
whau, nice
How many threads can 06 and Vantage utilize?Quote:
Originally Posted by s7e9h3n
3DMark Vantage only utilizes 8 threads for the CPU tests, I believe 3DMark 2006 also only utilizes 8 threads. So unfortunately I don't think this setup will produce monster scores like it is doing with wPrime. If those CPUs scaled like Deneb under LN2 a single chip would make for very interesting results.Quote:
Originally Posted by EnJoY
gomeler:--> it is not so good for Gulftown ,-] with 12 threads. Omg, i want see Thuban with LHe :D...more 1/2 year waiting...
me too.
AM3 compability is sweet :up:
one is running Sempron and can suddenly upgrade to Thuban :clap:
futuremarkQuote:
Originally Posted by [XC] gomeler
can't keep up with cpu future though lol
its good to see cpu score on 3dmark 06 or vantage .. vantage for sure is 16 thread cpu test ..
http://www.hwbot.org/hallOfFame.do?t...plicationId=15Quote:
Originally Posted by s7e9h3n
hwbot says you're third :up:
Are the multi's unlocked on the processors?
AFAIK, you can somehow 'disable' CnQ by setting all the P states to the same voltage and multi.
OMG did u people see this???
http://forums.ocworkbench.com/showthread.php?t=89809
http://img89.imageshack.us/img89/9111/obero.png
:rolleyes: :confused::shrug:
My dreams are true then L1 128 KbytesQuote:
Originally Posted by ajaidev
38-40MB Of cache O_o
42 secs on 1024m wprime it's 38% faster then mangus with only 25% more cores. at nearly the same clock speed, (3% lower clock speed actually.)
Don't meen to go off-topic, but didn't Intels new Gulftown score about 50% above the previous record in CPU Score?:shrug:Quote:
Originally Posted by [XC] gomeler
It seems too early for any Bulldozer leaks..
WTF???What is it? Orochi sample?Quote:
Originally Posted by ajaidev
edit:yes, Orochi ! Whau!!! AVX, 32nm, 2011 omg, hell CPU!
really...? (why) with Istanbul out ahead of time and mangus would ahead because that too.Quote:
Originally Posted by Lightman
so next step is bulldozer, I don't think it's too early. it's not going to be shelfs this year but maybe by mid nest year it gets there early from getting heavy testing make sure there aren't any bugs.
unlike that Barcelona which go pushed out without the heavy tested needed where the tlb bug was over hyped anyways.
Ahm, 256k.Quote:
Originally Posted by demonkevy666
The K7 already head 128k L1 cache...
right, still a monster though.Quote:
Originally Posted by Stefan Payne
This is very exciting. If I'm not mistaken, this is the first we've seen of early silicon. Considering the launch for this product is supposed to be well over a year from now, I hold this as a good sign.
yes, its really Bulldozer:eek::up:
right, to coming Bulldozer to servers is only 1 year maybe and to desktop last news (few days ago) talking about 4Q 2010Quote:
Originally Posted by demonkevy666
Makes sense that early silicone already is out...Quote:
Originally Posted by Lightman
That screenshot looks like a gold statue while ati 58## look like broken window.
28m l3 cache? Also according to cpu-z isn't it a multi-socket, 32 cores maybe?
lol, and here i was thinking its a mangus shrunk....
nice AVX catch!! But if its early silicon that means changes can be made i hope noting too drastic. Also if this is Bulldozer i am a bit sad on the performance 38% more with 33% more cores i was hoping for more...
EDIT Sorry wrong calculations....
CAN ANYONE TRANSLATE THIS:-
{
稍後將顯示更多的中國走出去
新藥實際上是到2011年 所以不能顯示太多
}
This was written under the pic!!!
25% more cores. 38% more 3% less clock so it can only go up actually.Quote:
Originally Posted by ajaidev
if it where 33% more cores then 24 core mangus
15% faster cores vs mangus.
if it where 25% more cores then 24 core mangus
52% faster cores vs mangus.
my math sucks lol
to get 67.424 or close to it I needed 1.58 X from 42.727.
its good for performance in wprime. Watch at Nehalem vs Westmere, clock for clock about zero up...
See what i found
http://images.hardware.info/news/opteron_april09b.jpg
So where does 16 core AMD fit, INTERLAGOS in 2011 sadly its MP but at least its G34 is INTERLAGOS suppose to be bulldozer???
I wonder if it has cmt like rumors would cpu-z still read it as 2 cores when its more like 2 threads?
Maybe SMT will be implemented later, this maybe a early silicon and can be modified to suite AMD's needs and preferences !!!
I thought 32nm for AMD is only available by 2H 2010! Interlagos should be Bulldozer. They promised Bulldozer by 2011.
Interlagos is bulldozer ;) This 16-core CPU is infact 2 Bulldozer(8-core) in MCM.Quote:
Originally Posted by ajaidev
salt anyone?
Me me, well i mean it can be the bulldozer and a early silicon that means that these specs are not air tight there will be most likely be changes brought at a later stage and thinking of one possible outcome as a final result is kinda not wise.Quote:
Originally Posted by Nedjo
So have a little salt and u will not be sad when bulldozer has 20 Mb of l3 instead of 28mb
That L3 size, 128KB L1's, AVX Instructions, 32nm... put some SMT technology and it's Bulldozer for me. Interlagos?
I think we should not highjack S7's thread. To discuss that create a separate thread please.
http://cpu.rightmark.org/download/rmma38bin.rar
The multi-threaded sub-test should run as many threads as possible,and
can do read/write per thread (custom size) -great for ganged-unganged modes.I am guessing there should be "minor" improvements in the
bandwidth "department" ;)
http://cpu.rightmark.org/download.shtml
Salt please.That website is a known one for faking results before. BTW it's too early for Interlagos samples,especially for the ones that supposedly ended up in hands of some Chinese PC enthusiast.
I suppose Steven's thread was a bit derailed.
Yes, please!Quote:
Originally Posted by Nedjo
I know that being pessimistic is less favoured here, but I always thought that this early in development (over 1 year before launch) if there is a working silicon, then it will be hidden deeply in AMD U.S. labs for internal validations. Not in OEM hands somewhere in far east.
Just look at Intel leaks, they are maximum 6 months before official product launches, that's when OEM's and ODM's are getting their first samples for validations.
I wish AMD would by this far ahead with Bulldozer, but I can't rationale that ATM.
use translator from japan to english and we will see... :) Think,its not fake
I want to see s7's 3d runs:up:
Its a fake
+1 yeah I still want to see 3dmark06 and vantage here too.Quote:
Originally Posted by charged3800z24
Sorry guys...I've got the system tied up for some "other" testing ATM ;) Benching will have to wait.
Looks like I missed a lot in the past few posts....I'll check with Franck and see what he thinks of that screeny....
we'll keep watching:up: and and thanks for checking on that screen shot too.
why do u mean? I have home screenshot Deneb from April 2008 in revision C0. --->+- one year before launch. I have confirmed exists samples Bulldozer. Why should it necessary fake?:down:Quote:
Originally Posted by w0mbat
16 core already :eek::eek:Quote:
Originally Posted by ajaidev
Don't bother, it's fake.Quote:
Originally Posted by s7e9h3n
http://www.xtremesystems.org/forums/...d.php?t=235030
Beat me to it:Quote:
Originally Posted by Piotrsama
Hi Steve,
1.52.3 does exist, it fixes several issues with Gulftown.
But that screen is a fake. cpuz does not know any 32nm AMD processor, so there is no chance that it displays that.
Franck
Now back to s7e9h3n's MagnyC. CPUs please :).
New Server / Workstation board and 2 of these CPU'S please:)
Ey, need some more info on that one, benches? :DQuote:
Originally Posted by s7e9h3n
yes, but now back to like Magn Course benching :)
There's some pictures of the kroner platform and a tyan board in this thread at AMDZone, as well as a picture of socket G34. There's a link to a new server blog also.
New AMD Server Platform Blog
how much power did the system need to run all cores at 3.2ghz? is it in any TDP range that desktops can reach with a 1P? i dont know why amd wouldnt want to make these for desktop just so they can sell for 1000$+ and laugh at SMT for a little while.
It's fake, directly from the horses mouth:Quote:
JF-AMD is actually John Fruehe, Director of Business Development for Server/Workstation products at AMD.Quote:
JF-AMD said: Here is how they accomplished that screenshot:
http://www.d3velopers.com/files/imag...shop-cs2_0.jpg
I can confidently say that it was not real.
Check it out here:
http://www.amdzone.com/phpbb3/viewto...&sd=a&start=25
That guy is good, I wonder how he knew it was made with CS2 and not CS4!
Because he's just that good:ROTF:Quote:
Originally Posted by mursaat
Honestly, the reason I even bother with AMDzone is because of John Freuhe stuff often "slips" out of hes mouth and he's good at handing out "clues" ^^ If it's intentional or not I am not sure :D
ANd what new with s7e9h3n benches?
someone send this man 5870 stat!
then we'll see if K10 is bottleneck on gpu hehe :D
Send him some more time also ;)Quote:
Originally Posted by demonkevy666
anyone else think s7 is the biggest tease on here? raise your hand if youve had blue balls for the last week!
(we do appreciate all the time youve given us, but theres not such thing as enough)
whats new man with Magny?
If my estimates are correct, this (2x12 core@3Ghz) MC system when running CMD2 WUs from BOINC would get about 90K PPD.
(Calculations based on Phenom 2 x4@ 3,33Ghz scaled to 3Ghz)
Avg PPD per core per day 3850
Times 24 = 92400
:shocked:
Edit: 8soc 2,2Ghz MC would get ~270K:eek: (about affordable in 3 years time)
Man, I must be slowing down.
New nice dualie toys and I didn't see it for almost 2 months..
Nice setup Steve..:up:
Took you some time to find this thread Dave! :rofl:Quote:
Originally Posted by Movieman
Nice toys indeed :D