Nehalem-EP......BLOOMFIELD

[Donnie27]

If that's a mobo teething issue, then you have to conclude that design changes in the mobo can affect the performance of the IMC.

I'd rather they tried 3 channel of DDR2 Hell, I can't wait to see what just two will do.

[Face]

BTW, it seems like the article missed a few screenshots or that the Author has updated the original post with more benchmarks.

You can see the original thread here:
http://bbs.chiphell.com/viewthread.p...a=page%3D1

WinRar


Sandra Proccessor Multi-Media


Sandra Multi-Core Efficiency



This benchmark does not test processor unit computational performance, i.e. how fast the cores of the processors are; it tests how fast is the connection between them only.

All very impressive, while the last one looks particularly good. IMC is looking great despite all the reported bugs etc..

[JumpingJack]

The winrar result is particularly impressive. The winrar internal benchmark is has always been sensitive to BW ... for comparision, a 2.6 GHz phenom scores ~1800 http://www.sharkyextreme.com/hardwar...261_3756681__5

[Shintai]

BTW, it seems like the article missed a few screenshots or that the Author has updated the original post with more benchmarks.

You can see the original thread here:
http://bbs.chiphell.com/viewthread.p...extra=page%3D1

WinRar


Sandra Proccessor Multi-Media


Sandra Multi-Core Efficiency



This benchmark does not test processor unit computational performance, i.e. how fast the cores of the processors are; it tests how fast is the connection between them only.

All very impressive, while the last one looks particularly good. IMC is looking great despite all the reported bugs etc..

Thats very very impressive

[Fuji]

Not true.

Lots of the early X58 boards are having trouble running Tri-channel memory, even though Nehalem has an IMC.

This bug WILL be corrected before retail boards ship, however.

i remember reading on fudzilla that it had to do with the trace length for the memory. Something along the lines of if there is extra delay in one of the channels, you have a rather large problem

[Shintai]

i remember reading on fudzilla that it had to do with the trace length for the memory. Something along the lines of if there is extra delay in one of the channels, you have a rather large problem

Considering one that already got a Nehalem CPU just says its a new BIOS to run 3 trichannel....then what you think

[-TFR-]

This extraction is not quite fair, nor correct. The entire memory sub-system for Nehalem has been reworked, you cannot look at the off-die DDR2-DDR3 comparision and make the assumption you are making.

In terms of marvels, who knows ... we have very little info to go by ... however, the hype machine of Geslinger compared the leap of Nehalem over Conroe to the magnitude of Conroe over Netburst, this indeed would be marvelous. My take on it is that it will be less overwhelming in single thread, but multithreaded situations will be impressive on this magnitude... Anand's preview data seems to indicate that this is indeed the potential... 30-50% gains depending on the app (multithreaded of course).

This will be even more impressive in server, where the mem BW issue is much much more pronounced... Kanter (RWT) seems to think Nehalem will be nothing short of a miracle. Time will tell.... however, initial indications are very positive that your downplay of Nehalem is probably not warranted.

Jack

This article shows how much a K10 cpu benefits from increased bandwidth, i know nehalem is another arch, but still an interesting read
http://www.digit-life.com/articles3/...r2-533-p1.html

[BulldogPO]

Why anobody has´nt shown us Sandra or Everest memory bandwidth tests?

[Movieman]

-TFR- was Gothr3K..


On the rest of this I'm afraid we'll just have to wait to see what Nehalem does with a board with working trichannel memory..
There's an old line:
The proof of the pudding is in the eating.. and you don't judge the pudding until it's totally cooked.

[MomijiTMO]

Mwwahahhaha some people can't help themselves.

[Movieman]

Mwwahahhaha some people can't help themselves.

I sent him this and putting it here so there is a datestamped public record of it:
Niels;
This multiple account crap is going to stop.
I will be very frank with you.
If you continue I will take two measures.
First I will contact your ISP as I have that info and do whatever is necessary to have your account pulled for harassing an internet web site(XS)
Second I will contact a member here who works for Interpol and have him do what he can to make your life interesting.
As before, the choice is yours but you've now turned a 7 day ban into a personal thing.
If you want a game out of this, please just continue pushing.
I have the time and the contacts to play this out.
It's all up to you.

[Hornet331]

On the rest of this I'm afraid we'll just have to wait to see what Nehalem does with a board with working trichannel memory..
There's an old line:
The proof of the pudding is in the eating.. and you don't judge the pudding until it's totally cooked.

indeed, but this preliminary results are already good. So i doubt we can see anything worse, unless Intel screws something big up.

[tiro_uspsss]

indeed, but this preliminary results are already good. So i doubt we can see anything worse, unless Intel screws something big up.

its like dipping ur finger into the ooh soo sweet mixture

[tiro_uspsss]

thats it - who mentioned food - i'm hungry now!

Movieman started it! Ah-ha!

now look what u did!

[Movieman]

indeed, but this preliminary results are already good. So i doubt we can see anything worse, unless Intel screws something big up.

I agree totally.. Can't wait..
Just some observations;
The only 45nm chips I have are in the dual socket SuperMicro board here.
E5450 quadcores at 3404mhz(3000/12mb/1333 stock)(Harpertown's)
I do DC work with the systems here.
2-8 core clovertown's at 3000, the 8 core Harpertown and a Q6600 at 3600
In WCG the 4 core Q6600 does app 15,000 points a day
The 8 core clovers do app 25-26,000 a day
The Harpertown is averaging app 34,000 points a day
Lets assume there's a 5-10% improvement with Nehalem in single threaded apps and a 30% in apps that can take advantage of this new HT.
Lets also assume that Nehalem will be able to reach at least 3400 on air.
An 8 core Nehalem(Gainstown) system will be able to work 16 work units at a time vs the Clover and Harper's 8 at a time and in doing so will bring into use this new HT with it's presumed 30% increase clock for clock.
Now you see why I am so excited about this.

[Hornet331]

I agree totally.. Can't wait..
Just some observations;
The only 45nm chips I have are in the dual socket SuperMicro board here.
E5450 quadcores at 3404mhz(3000/12mb/1333 stock)(Harpertown's)
I do DC work with the systems here.
2-8 core clovertown's at 3000, the 8 core Harpertown and a Q6600 at 3600
In WCG the 4 core Q6600 does app 15,000 points a day
The 8 core clovers do app 25-26,000 a day
The Harpertown is averaging app 34,000 points a day
Lets assume there's a 5-10% improvement with Nehalem in single threaded apps and a 30% in apps that can take advantage of this new HT.
Lets also assume that Nehalem will be able to reach at least 3400 on air.
An 8 core Nehalem(Gainstown) system will be able to work 16 work units at a time vs the Clover and Harper's 8 at a time and in doing so will bring into use this new HT with it's presumed 30% increase clock for clock.
Now you see why I am so excited about this.

indeed, for myself i also do some boinc crunching (seti, rosetta), but only when my computer is on (~8h a day). Its not much but it still helps, plus doing 8 WU at once compared to 4, will also help my RAC.

[Movieman]

indeed, for myself i also do some boinc crunching (seti, rosetta), but only when my computer is on (~8h a day). Its not much but it still helps, plus doing 8 WU at once compared to 4, will also help my RAC.

We need to get you over to WCG..

[mstp2009]

-TFR- was Gothr3K..

You guys should do an IP ban. Works wonders for us.

[Neuuubeh]

Looking nice. Kinda gets you thinking about them prices tho .

[Hornet331]

We need to get you over to WCG..

maybe with nehalem, i could spare 3 threads.

[Movieman]

maybe with nehalem, i could spare 3 threads.

Just 3? Where's my ban hammer!

[Trunks]

You guys should do an IP ban. Works wonders for us.

IP ban is useless against dynamic IP.

[Movieman]

IP ban is useless against dynamic IP.

True, this is where you have his ISP block OUR IP from his account.

[Shintai]

True, this is where you have his ISP block OUR IP from his account.

Be happy you dont contact me at work (ISP). We tend to block nothing unless there is a courtorder

But anyway...there would be proxies...

[Hornet331]

Just 3? Where's my ban hammer!

uh noes!!!

but i think about switching from rosetta to Artificial Intelligence.