K10 demo machine in Akiba

Printable View

09-13-2007, 04:59 AM
kyosen

K10 demo machine in Akiba

As PC Watch reported yesterday, one PC shop "USER SIDE" in Akiba has started
K10 demo with Tyan n3600M + Opteron 2347 x2.
http://www.watch.impress.co.jp/akiba...elonademo.html

Today, I have a chance to touch the demo machine.
It's configured as virtual machine with VMWare, hosted by Linux.
Here is a preliminary test result on virtualized WinXP.

http://222.151.152.227/c-board/file/...47_CM_SP1M.png

very sloooow SuperPI time!?...
...yeah, it's nearly same as K8 by clock for clock, it looks.

I'll try SuperPI again on pure Server2003 in next time,
but I won't get any drastic improvement, I expect.
Also I may have a chance to try 3DMark...anyway, interesting:)
09-13-2007, 05:03 AM
stealth17

damn that sucks. All we can hope is that these things overclock like a scolded dog (no cold bug would be sick!)
09-13-2007, 05:07 AM
Okda

overclocking won't be enough if those numbers are true which i hope not
09-13-2007, 05:09 AM
LIKMARK

This one is of the B1 revision?
09-13-2007, 05:10 AM
Motiv

Quote:

Originally Posted by kyosen

http://222.151.152.227/c-board/file/...47_CM_SP1M.png

very sloooow SuperPI time!?...
...yeah, it's nearly same as K8 by clock for clock, it looks.

I'll try SuperPI again on pure Server2003 in next time,
but I won't get any drastic improvement, I expect.
Also I may have a chance to try 3DMark...anyway, interesting:)

Is that Super Pi running in VMware? if so, what kind of performance degradation would it cause?
09-13-2007, 05:15 AM
Sparky

We all know that K10 isn't as good as intel in superpi. No surprise here. But superpi is not a good indicator of overall performance and I don't give a flying crap about it much really.
09-13-2007, 05:20 AM
Motiv

Quote:

Originally Posted by SparkyJJO

We all know that K10 isn't as good as intel in superpi. No surprise here. But superpi is not a good indicator of overall performance and I don't give a flying crap about it much really.

I like it as an overclocking Benchmark. It gives you a nice indication of how much better your fine tuning is going to get that last 0.01% increase from your CPU.

It also gives a decent indication of how much better the same platform increases with newer revisions of the same core.
09-13-2007, 05:28 AM
Sparky

Quote:

Originally Posted by Motiv

I like it as an overclocking Benchmark. It gives you a nice indication of how much better your fine tuning is going to get that last 0.01% increase from your CPU.

It also gives a decent indication of how much better the same platform increases with newer revisions of the same core.

This is true.

IIRC, intel CPUs have tended to be faster at superpi anyway haven't they?
09-13-2007, 05:35 AM
hardrail

A machine that is running a virtualized OS on VMWare will always be a pig...

Still waiting on a solid benchmark with a freshly installed physical OS.
09-13-2007, 05:54 AM
funnyperson1

Quote:

Originally Posted by SparkyJJO

This is true.

IIRC, intel CPUs have tended to be faster at superpi anyway haven't they?

Netburst was worse than K8, but Dothan was better. The C2Ds are disproportionately better than K8 of course compared to real world applications.
09-13-2007, 06:09 AM
BeardyMan

Superpi IS L2 cache benchmark.

give the barca 4MB L2 cache and see the enormous pi boost...
09-13-2007, 06:22 AM
Bail_w

Not impressed with the results, and i dont think the overclocking will fly off the chart.
09-13-2007, 06:25 AM
Donnie27

Quote:

Originally Posted by BeardyMan

Superpi IS L2 cache benchmark.

give the barca 4MB L2 cache and see the enormous pi boost...

So we should wait for Barkie to get a 4MB L3? Or do you mean L2?
09-13-2007, 06:50 AM
macci

Run 3DMark05 instead ;)

SuperPi doesnt like AMD..
09-13-2007, 07:07 AM
tux08902

If you're running it in VMWare, then you can expect a drastic decrease in performance.
09-13-2007, 07:15 AM
planoru

Something is wrong. That's the result I got on a 1700+@2100(11xFSB200). This can't be. It's way slower then any k8@2GHz.
09-13-2007, 07:16 AM
Kobalt

1 Attachment(s)

Here is K8 at 1.9GHz, 190fsb, 512mb of DDR at 2-2-2-6. I'd say VMWare slows it down quite a bit.
09-13-2007, 07:19 AM
Start

2 Attachment(s)

I wish I had an AMD rig to test it, but I don't anymore :(. I do however, have Intel Core Duo (Yonah) @ 2 Ghz. Left screenshot is Parallels 3. Right is Vmware Fusion. Both 1M run.
09-13-2007, 08:16 AM
kyosen

LIKMARK> This one is of the B1 revision?
--
Yeah, I think so...at least, it shows "1" as stepping value.

I had no experience of VMWare/virtulization at all till today,
so I have no idea for degree of performance degradation...

BTW I have no interest for such an argument whether SuperPI is
suitable for Quad-core CPU performance evaluation or not...
I do it just to fulfill my curiosity.
Also I know supreme advantage of Intel Core2 for SuperPI, ENOUGH;)
So go on topic, and stay tuned...

Oh macci, thanks for giving your glance at my post!!!
I also wonder how much 3DMark06's CPU score goes:)
09-13-2007, 08:22 AM
Eldonko

Dam, I was hoping for much better PI time than that. :(
09-13-2007, 09:28 AM
PhilDoc

Thanks for sharing your info, Kyosen, looking forward to anything you have to show us. :up:
09-13-2007, 10:00 AM
Lightman

Thanks for sharing Kyosen! :up:

Are you planning second trip to this shop? Can you run Cinebench 9.5 x64 on that system?
Regarding SuperPi result it is low due to test platform. It is a server in the end with Reg. DDR2-667 and we aren't sure if NUMA is set properly, so in case it isn't performance decrease is huge...
09-13-2007, 10:08 AM
Jacky

nice results :)

However I hear so much talk about K10/K8 being superior on virtualization to anything Intel can offer. so much talk, but no facts. we don't even know what the performance decrease of runing in a VM could be.. anyway it should be lower than with intel, it should... :(

It would be great if someone could run a conroe w/ and w/o a VM and then compare the perf. decrease to get an idea. maybe someone that believes in AMD's superiority in virtualization like nn_step :D
09-13-2007, 10:29 AM
Sampsa

VMWare/virtulization will definitely cause penalty speed, how much? I don't know.
09-13-2007, 10:43 AM
massman

Quote:

Originally Posted by macci

Run 3DMark05 instead ;)

SuperPi doesnt like AMD..

C'mon, we want to see 3DMark01 :D

Thanks for sharing Kyosen
09-13-2007, 10:47 AM
Vapor

I can easily get performance within a scant 2-4% on SuperPi with virtualizing Windows on top of OS X...this is untweaked, just install and run. This run does look very slow though.

Expect K10's performance product to be at or above 71k all said and done from what I've seen :( (C2D at just over 50k)

There's still hope in 3D, but competing with 5+GHz Kentsfields and 5.5GHz Conroes will be....not easy

Maybe for 24/7 they'll be good too, less worry about a feared coldbug :)
09-13-2007, 10:55 AM
Dagalidis

All those Benches are a BAD JOKE...:down:

Please don't waste your time on a simulated OS...
Run again on pure OS and post again....;)
09-13-2007, 11:04 AM
caligula

You lose no more than 5% with VMWare. I dunno about Xen (err... I mean qemu ;)). Overall, the biggest impact will be no more than 10%....
09-13-2007, 11:17 AM
duploxxx

Quote:

Originally Posted by caligula

You lose no more than 5% with VMWare. I dunno about Xen (err... I mean qemu ;)). Overall, the biggest impact will be no more than 10%....

it will depend. if it is VMware esx you could say about 5%, if it is vmware server (gsx) or workstation it is even worse.

so i would say. nice you have one, but stop posting results until you are at os level.

or if you can bench on ESX 3i, i would love to see the vmmark score :)
09-13-2007, 11:17 AM
informal

Guys look at s7e9h3n's results on his K10 ES on dually board:
http://www.xtremesystems.org/forums/...&postcount=335

According to his tests,the CPU(due to a bios bug) was running@800Mhz with 160Mhz for the ram and scored 39s in SPi 1M :D.A lot better than Akiba system @2Ghz
09-13-2007, 11:41 AM
gone_fishin

Wow, I never seen so much grasping at straws in my life.
09-13-2007, 11:55 AM
informal

Quote:

Originally Posted by gone_fishin

Wow, I never seen so much grasping at straws in my life.

Like Perkam said in another thread,intel section is a few cm down from the AMD one,so feel free to go there and stop watching this grasping at straws anymore:rolleyes:
09-13-2007, 12:00 PM
rogueagent6

Quote:

Originally Posted by gone_fishin

Wow, I never seen so much grasping at straws in my life.

Crap much? :rolleyes:
09-13-2007, 12:01 PM
gone_fishin

Quote:

Originally Posted by informal

Like Perkam said in another thread,intel section is a few cm down from the AMD one,so feel free to go there and stop watching this grasping at straws anymore:rolleyes:

I am a potential customer, not some mindless fanboy.
I want to see real numbers that are proudly presented to the public to demonstrate the benefits of a new product. And I want them to be independantly verifiable.
09-13-2007, 12:04 PM
ownage

50 seconds superpi 1M cant be true.
Not possible.

I have winXP virtualized on a 2.8ghz X2 and it scores much better.

http://img410.imageshack.us/img410/8...wnedwn7.th.jpg
09-13-2007, 12:05 PM
PhilDoc

Quote:

Originally Posted by gone_fishin

I am a potential customer, not some mindless fanboy.
I want to see real numbers that are proudly presented to the public to demonstrate the benefits of a new product. And I want them to be independantly verifiable.

No one is asking anyone to be a fanboy, just respectful. I think we have enough of this in other threads. Lets try to keep this one clean.
09-13-2007, 12:08 PM
ownage

I know 2.8ghz is something differend then 1.9ghz, but K10 is something new and i would like to see much better clock-for-clock performance.

But my virtualized superpi time is almost the same as my native winxp superpi time, so i dont know if it is virtualisation which slows it down.
09-13-2007, 12:09 PM
gone_fishin

Quote:

Originally Posted by PhilDoc

No one is asking anyone to be a fanboy, just respectful. I think we have enough of this in other threads. Lets try to keep this one clean.

I have recieved nothing but disrespectful replies.
This thread has no real value when the numbers presented are not believed and wild interpretations of them ensue. I only pointed out the obvious, grasping at straws.
09-13-2007, 01:41 PM
xVeinx

My AMD 4600+ X2 does PI faster than that...

http://img.photobucket.com/albums/v2...n/screenie.jpg
09-13-2007, 02:07 PM
showmanlkz

That cant be true? Should try to open 4 PI rather than 1, the result should be interesting....
09-13-2007, 06:16 PM
cky2k6

Quote:

Originally Posted by showmanlkz

That cant be true? Should try to open 4 PI rather than 1, the result should be interesting....

yep, as s7e9h3n said in his thread, the cpus fumble around the load to each other, maybe filling them all up would reduce this effect.
09-13-2007, 06:33 PM
guess2098

try a software is "NOT" depends on frequence but support multi core :)
09-13-2007, 07:14 PM
GAR

super pi doesnt mean JACK
09-13-2007, 09:36 PM
nn_step

Quote:

Originally Posted by Jacky

nice results :)

However I hear so much talk about K10/K8 being superior on virtualization to anything Intel can offer. so much talk, but no facts. we don't even know what the performance decrease of runing in a VM could be.. anyway it should be lower than with intel, it should... :(

It would be great if someone could run a conroe w/ and w/o a VM and then compare the perf. decrease to get an idea. maybe someone that believes in AMD's superiority in virtualization like nn_step :D

I'll tell you flat out, the performance hit for Intel for virtualization will about about 10 times that AMD experiences. (simply because their Hypervisor in via software verse AMD's hardware)
09-13-2007, 09:41 PM
JumpingJack

Quote:

Originally Posted by SparkyJJO

This is true.

IIRC, intel CPUs have tended to be faster at superpi anyway haven't they?

Prescott was not that fast :) .... I can show benches if you like. :)
09-13-2007, 09:58 PM
PhilDoc

Seems to me that FUGGER got some pretty good spi numbers out of Prescott, but it put out a ton of heat. Not sure what the final results were. It was about that time I got burned out a little on OCing and stayed away until just recently
09-13-2007, 10:02 PM
nn_step

Quote:

Originally Posted by JumpingJack

Prescott was not that fast :) .... I can show benches if you like. :)

SuperPI tends to prefer large Caches and short pipelines.
AMD had 1MB L2 when Northwood only had 512Kb
AMD also had shorter pipelines but now the difference is significantly less.
09-14-2007, 08:01 AM
kyosen

Trial with pure Server2003

Today I brought my HDD to the shop, and got the chance to check booting
display and BIOS setting.

S2932, with latest BIOS 2.01 on Tyan site, detected 1.9GHz core x8 properly.
it also detected the North-Bridge as Rev. B1
http://222.151.152.227/c-board/file/S2932_boot-up.jpg

The North-Bridge is running at 1400MHz.
It seems that FSB(=HyperTransport, I think) multiplier is restricted upto x4.
http://222.151.152.227/c-board/file/...IOS_NB1.4G.jpg

SuperPI 1M time = 42.781s ...
...yeah, it's improved from yesterday(on virtualized WinXP).
BTW, CPU-Z shows that the system is running at Single channle mode:(
http://222.151.152.227/c-board/file/...32-Opt2347.png

By K8, 52.204@1.9G=200x9.5 result, at nearly same memory-timing:
http://222.151.152.227/c-board/file/...64_SC_2CPU.png

Also by K8, 41.484s@2.0G=200x10.0 result...
http://222.151.152.227/c-board/file/...64_SC_2CPU.png
So, in K8, the difference between x9.5 and x10.0 is larger than simple clock ratio.
I want to check K10 behaviour @2.0G=200x10 or @1.8G=200x9...

...Good night:)
09-14-2007, 09:34 AM
freeloader

I've asked before but didn't get an answer. In this situation, how come the memory controller is running slower that the CPU? 1400mhz MC vs 1900mhz CPU. I thought the MC would be running at the same speed as the CPU.
09-14-2007, 09:40 AM
metro.cl

Kyosen are you sure that CPU-z is working OK might be just a problem readying.
09-14-2007, 09:48 AM
PhilDoc

Quote:

Originally Posted by freeloader

I've asked before but didn't get an answer. In this situation, how come the memory controller is running slower that the CPU? 1400mhz MC vs 1900mhz CPU. I thought the MC would be running at the same speed as the CPU.

I think Dave was working on this. He was saying 400 for single power plane and 200 for dual. He was also saying something about a bios change helping this, but it may take a cpu revision. He's the expert and would know. For me I wonder if Phenom would be the same. Ether way, getting them clock for clock would certanly help performance.
09-14-2007, 10:52 AM
masterg

if superpi runs like other programs like maya, there was a report done with quads i had it favorited but it wont come up anymore

that the quad has so much room with all those cores that the programs running on the quad basically freaks out and slows down if not stops because it doesnt know how to work with that much room, thats why the quads have worse scores then duals like in rendering and stuff
09-14-2007, 12:31 PM
kyosen

Quote:

Originally Posted by metro.cl

Kyosen are you sure that CPU-z is working OK might be just a problem readying.

Not sure...
...I should have checked it by decreasing memory modules, but not yet.
09-14-2007, 12:59 PM
GenTarkin

So.....Is that POST screen accurate in stating the cpu is a B1 revision?
Cuz, I know there is a lot of confusion going around about who is testing what revisions in reality. So, Im wondering is this truely the reflection of the B1 performance? ----meaning is this really a B1 or is the board confused and we still dont know...
Just wondering lol
09-14-2007, 01:06 PM
PhilDoc

From what I know all reviewer got B1 revisions, which are cripped ~5%, don't know exact percentage. The true shipping revision is suppose to be BA, which is not crippled.
09-14-2007, 01:25 PM
metro.cl

Quote:

Originally Posted by kyosen

Not sure...
...I should have checked it by decreasing memory modules, but not yet.

maybe try a diferent app just to see what happens
09-14-2007, 03:42 PM
GenTarkin

wait wait wait....whoa...I did not know that but only a 100mhz difference on a K8 yeilds 10 roughly ten seconds in SuperPI 1M?? @ that low of clock speed
Thats some crazy scaling.....
So, if the K10 scales anything like the few experiences have said...it does...then
The results of a k10 higher clocked should be some quite nice numbers no???
09-14-2007, 05:56 PM
KTE

Quote:

Originally Posted by LIKMARK

This one is of the B1 revision?

Yep, its B1. ;)

Kyosen: thanks bro. Your pic wasn't showing up to me y'sterday. :)

FWIW, thats exactly double the single core P4 3400MHz total score there. BUT the memory bandwidth is way lower, although the cache bandwidth is far higher.

Why do the cores read 900MHz bootup frequency and 111.25% oc? Are you sure they were all running at 1900MHz? :confused:
09-16-2007, 11:25 AM
s7e9h3n

Quote:

Originally Posted by kyosen

The North-Bridge is running at 1400MHz.
It seems that FSB(=HyperTransport, I think) multiplier is restricted upto x4.
http://222.151.152.227/c-board/file/...IOS_NB1.4G.jpg

The NB/IMC is running @ 1400mhz and not the rated 1600mhz because the motherboard either doesn't support dual power planes or the feature isn't enabled. If it was, the NB/IMC could run at spec'ed speed.

Quote:

Originally Posted by KTE

Why do the cores read 900MHz bootup frequency and 111.25% oc? Are you sure they were all running at 1900MHz? :confused:

The program is probably detecting the multiplier for the wrong p-state for the cpu.
09-18-2007, 03:45 AM
BeardyMan

Quote:

Originally Posted by macci

Run 3DMark05 instead ;)

SuperPi doesnt like AMD..

Marking your words here. :)

;)
09-18-2007, 03:52 AM
Equil|briuM

Wow, another pointless test.

How about testing it on -

A - A Fresh XP install.

B - FULL compatible hardware in full cooperation with K10.
09-18-2007, 04:12 AM
tictac

hei... the NB speed does it equal to LDT speed?
09-18-2007, 07:24 AM
rfurman24

Why not just use superpi for linux? As stated before just another worthless test.
09-19-2007, 12:06 AM
kyosen

Quote:

Originally Posted by tictac

hei... the NB speed does it equal to LDT speed?

No, the NB speed is 1.4GHz and LDT speed is 800MHz = 200MHz x4 in this boards, it looks.
09-19-2007, 06:19 AM
tictac

can we overclock nb speed? and if we can does it improve l3 cache or memory performance?
09-19-2007, 06:38 AM
dinos22

ah nice
kyosen raiding a joint with his HDD hahah fantastic
you sir are a true geek :D :up:
09-19-2007, 07:25 AM
informal

Quote:

Originally Posted by tictac

can we overclock nb speed? and if we can does it improve l3 cache or memory performance?

With Phenom it will be possible,and it will give a new dimension on AMD OCing :).Separate PLLs per core will make possible individual core OCing and separate power plane for IMC will make possible the higher freq. for IMC and L3 cache with independent VddNB adjustments.December can't come sooner enough :D
09-19-2007, 07:55 AM
erwinz

go kyosen :D
09-19-2007, 07:56 AM
Thesavage

Remember reading on the old aceshardware that a small oc on the L3 would provide a nice speed bump, not sure if that is true though.
09-19-2007, 06:48 PM
macci

tictac,

HT speed can be equal to NB speed w/ HT3 chipsets..

overclocking the NB will result in improved memory performance. Can be overclocked via HT ref. clock or via NB multiplier (which may or may not be locked..).
09-19-2007, 10:01 PM
KTE

Lower L3 cache latency is one benefit of that.
09-19-2007, 10:20 PM
IvanAndreevich

I can't wait to see AM2+ samples tested in the wild. These crippled mobos / bios / old rev core (which apparently possibly doesn't have L2 working or something along those lines) don't mean much. But this DOES SHOW how far AMD is from actually launching this CPU on the market.

macci
Don't they want to throw some K10 AM2 CPUs your way to put on the autocascade and prove that AMD is still in the game? I think the PR stunt Intel got FUGGER to do for them by breaking WRs with Penryn is kickass.
09-20-2007, 08:11 AM
GAR

Super Pi is a cache test, the more cache the faster it will go, it doesnt mean its going to perform less in games, super pi doesnt mean jack.
09-20-2007, 08:20 AM
tictac

wow.. thx macci.. that would be great.. so we need to wait for rd790 board featuring ht3. im getting more confused lately with all this new stuff nb speed , l3 cache latency.. ht3 speed.. :p
09-20-2007, 08:23 AM
BeardyMan

Quote:

Originally Posted by GAR

Super Pi is a cache test, the more cache the faster it will go, it doesnt mean its going to perform less in games, super pi doesnt mean jack.

I can't disagree but,
Superpi 1M is like the 100M sprint men in the athletics , queens sprint.
the ultimate speed test..
09-20-2007, 08:40 AM
tictac

hei.. if the nb speed are not connected to cpu speed.. the memory speed wont change if we overclock the htt speed right.. so imagine 1.9ghz processor with only 1.4ghz nb with memory ratio set at 200mhz that will only give 155mhz memory speed.. so the memory speed is underpower right?
09-20-2007, 09:00 AM
doompc

MEM Clock is still connected to CPU Clock:
http://222.151.152.227/c-board/file/...32-Opt2347.png

What isn't clear is where the NB Clock come from.
There might be another multiplier setting.
(lol 3 multipliers: CPU, HTT and now NB)
09-20-2007, 12:55 PM
Bigchrome

so almost 10 seconds slower than my 4 year old mid range A64....That's good improvements anyway.

:banana::banana::banana::banana:ing NOT!
09-20-2007, 02:14 PM
SOLDNER-MOFO64

Quote:

Originally Posted by Bigchrome

so almost 10 seconds slower than my 4 year old mid range A64....That's good improvements anyway.

:banana::banana::banana::banana:ing NOT!

If even for a second you, or anyone else for that matter imagine this scenario to be correct then you need serious help.
09-20-2007, 02:18 PM
SnipingWaste

I think ppl are forgetting that AMD ES CPUs are not as good as the retail CPU's. Look at the A64 benches with the 3100+ A64 ES CPU's that ran at 1.8gig with 1 meg L2. It could only run with T2 and not T1 and the memory controller would error above DDR333. The retail 3200+(2.0gig) running at 1.8gig out performed the 3100+ and I think we will see the same thing with the K10. It would be nice to see some benches off a retail K10 BA or up stepping and see the performance different s.
09-20-2007, 02:57 PM
KTE

Quote:

Originally Posted by SOLDNER-MOFO64

If even for a second you, or anyone else for that matter imagine this scenario to be correct then you need serious help.

SOLDNER... SPI doesn't bother me but what do you think K10 1.9GHz or 2GHz time will be for a 1M?
09-20-2007, 03:08 PM
kyosen

Quote:

Originally Posted by SOLDNER-MOFO64

If even for a second you, or anyone else for that matter imagine this scenario to be correct then you need serious help.

SOLDNER... SPI does bother me then what do you think K10 1.9GHz or 2GHz time will be for a 1M?:)
09-20-2007, 03:09 PM
leoftw

Why test it on Vmware ? Thats a horrible idea IMHO .

Here are my quad opteron tests

On 32bit Vista and 2.6 Quad Optys , 4gb DDR2 667 ECC

http://i43.photobucket.com/albums/e3...titled-1-1.jpg
09-20-2007, 03:46 PM
knightwolf654

those are duel cores....
09-20-2007, 03:57 PM
macci

NB (inside the CPU, the memcontroller) has its own multiplier
CPU has its own multiplier (for each core that is..)
HT bus has its own multiplier (which can not exceed the NB multiplier)

Memory is connected to HT ref. clock (but not to overall CPU clock anymore), you have 200,266,333,400 and 533MHz... memclk modes.

Memory clock goes up when you adjust HT ref. clock (e.g. if you boot at 200 ref. clock / 400MHz memclk and then increase the ref clock to 205MHz your memclk will be 410MHz (DDR2-820)

since the memclk is driven by the ht ref.clock we dont get wierd memclks with certain multipliers (e.g. 375MHz with x15 in case of K8)

memory speed readings w/ CPU-Z (at least ver 1.41) were probably incorrect (as it assumed memclk is still done the K8-way)
09-20-2007, 04:51 PM
KTE

Quote:

Originally Posted by kyosen

SOLDNER... SPI does bother me then what do you think K10 1.9GHz or 2GHz time will be for a 1M?:)

.... :D
09-20-2007, 06:13 PM
doompc

@ macci, with MEM Clock driven by HT Ref Clock we have 1:2, 3:4, 2:3, etc dividers again?
09-20-2007, 11:36 PM
leoftw

Quote:

Originally Posted by knightwolf654

those are duel cores....

its two dual cores , not one :D
09-20-2007, 11:53 PM
Motiv

Quote:

Originally Posted by KTE

SOLDNER... SPI doesn't bother me but what do you think K10 1.9GHz or 2GHz time will be for a 1M?

Quote:

Originally Posted by SOLDNER-MOFO64

From the benches I've been priveledged enough to see the results of (:))a 2ghz K10 will score around a 19sec 1M Pi time......at 3ghz approx 12secs.

:up:

Think he may of been of with his privileged information. Making claims like that always comes back to bite you on ya ass...

http://www.xtremesystems.org/forums/...&postcount=525
09-21-2007, 12:57 AM
KTE

Hmm.. well only he can answer what he saw and didn't. ;)

The only idea I have so far is by what the guys around online have done: ~40-49 sec at 1900MHz. Seems high but the CPUs should be out soon and we'll know for sure the rough range.
09-21-2007, 05:00 AM
tictac

Quote:

Originally Posted by macci

NB (inside the CPU, the memcontroller) has its own multiplier
CPU has its own multiplier (for each core that is..)
HT bus has its own multiplier (which can not exceed the NB multiplier)

Memory is connected to HT ref. clock (but not to overall CPU clock anymore), you have 200,266,333,400 and 533MHz... memclk modes.

Memory clock goes up when you adjust HT ref. clock (e.g. if you boot at 200 ref. clock / 400MHz memclk and then increase the ref clock to 205MHz your memclk will be 410MHz (DDR2-820)

since the memclk is driven by the ht ref.clock we dont get wierd memclks with certain multipliers (e.g. 375MHz with x15 in case of K8)

memory speed readings w/ CPU-Z (at least ver 1.41) were probably incorrect (as it assumed memclk is still done the K8-way)

Ok now we wont have weird memory speed with the new architecture... :D

Hei... are you hidding something from us? :D
09-22-2007, 03:35 PM
Bigchrome

Quote:

Originally Posted by SOLDNER-MOFO64

If even for a second you, or anyone else for that matter imagine this scenario to be correct then you need serious help.

So the results we've Just ACTUALLY SEEN are false? Wheras the ones you claim but have no way of verifying are more believeable. Sorry, I forgot my place for a second.:ROTF:
09-22-2007, 03:56 PM
uwackme

hmmm

Macci, is that true.... the cores have seperate multipliers?

Could be cool, keep temps lower on apps that only use one core, etc.
09-22-2007, 05:27 PM
LowRun

Quote:

Originally Posted by uwackme

Macci, is that true.... the cores have seperate multipliers?

Could be cool, keep temps lower on apps that only use one core, etc.

Hehe, dude this is Macci speaking, makes it as reliable as can be :)

All times are GMT -8. The time now is 06:03 PM.

XtremeSystems