Nehalem-EP......BLOOMFIELD

[XS Janus]

Any reason why I should buy kitted RAM vs. regular OEM sticks if I don't plan of OC the RAM?

And one more Q: The difference between CL7 and CL7 DDR3 1600 here is like 30% in price (cca 75$ more). How much of an actual real life system performance do you guys think we shall get from choosing CL7 over the cheaper CL9 memory?

[XS Janus]

Oh, OH!
OBR you said that the power consumption is huge! How so? How much vs. the system you compared it to? And what was the comparing machine built of?

[Shintai]

Any reason why I should buy kitted RAM vs. regular OEM sticks if I don't plan of OC the RAM?

And one more Q: The difference between CL7 and CL7 DDR3 1600 here is like 30% in price (cca 75$ more). How much of an actual real life system performance do you guys think we shall get from choosing CL7 over the cheaper CL9 memory?

Absolutely none.

[JumpingJack]

Nehalem is a monster, performance in some programs are shocking ... but especially in games is Core 2 still great (and mostly better) choice ...

The results should be mixed overall --- like you said, single threaded games will not see much, if any, and could actually be lower -- can you make a statement comparing a a single threaded game to a multithreaded game at resolutions (and using a GPU) that checks the CPU capabilities?

Example... say FEAR vs UT3?

[Drwho?]

I wonder if we'll have 3x2gb kits around launch time.

all the big memory players are doing it, by 3 packaging are ready, i was showing them at IDF.

[Linchpin]

all the big memory players are doing it, by 3 packaging are ready, i was showing them at IDF.

Awesome, thank you

[Hornet331]

I think you guys looking for raw speed( Da benchas) will be better off with 3x1 gig sticks and leave the 3x2 gig sticks to the guys for mainstream use.
I'm just looking at the cost of 6x1 gig for the dual boards and cringing..

well since on wu of rosetta can take up to 500mb or ram and with nehalem it can use up 8 threads -> thats 4 gig alone for rosetta

count in windwos vista, and a game and your beyond 6 gig.

[Movieman]

well since on wu of rosetta can take up to 500mb or ram and with nehalem it can use up 8 threads -> thats 4 gig alone for rosetta

count in windwos vista, and a game and your beyond 6 gig.

I don't do rosetta anymore.
That POS Baker doesn't deserve help from anyone.
When you bust your balls to help someone and then they spit on you it's time to leave and most of us said goodbye to that ass.
Sorry, but thats the truth.
I do WCG and 6 gig will be more than plenty to do 16WU at a time on a dual socket Nehalem.

[Shadowmage]

The cache and prefetchers makes memory latency almost irrelevant.

That's 100% not true.... everything in addition to an in-order, simple pipeline is to combat memory latency. That's the whole purpose of out-of-order pipelines: to find work to do while waiting for a load to return.


EDIT: To prevent people from nitpicking me: yes, OoO processing was first invented by Tomasulo at IBM to combat long FP latencies, but now load instructions are the main reason why OoO is still mainstream. If you have ever designed an OoO processor, you would know that the overhead is massive compared to a nice and simple in-order processor.

[FlanK3r]

Ideal server CPU and for MTH

[XS Janus]

Absolutely none.

Thanks Shintai!
I had a feeling that kited RAM and low CL was even less important than in DDR667 days.

[Anemone]

I don't think Nehalem will be another Prescott. But it seems strange to have crippled the L2 by such a large amount. L3 shared on all 4 cores is good, but certainly the value the L2 was playing was known. It's going to certainly be interesting to see where the i7 does well and where it does not before pushing the "buy" button...

[Mr.BSEL]

Over at pc games hardware there was an article about overclocking Nehalem to 4 ghz with an extreme processor but it took 1.7 volts to reach that speed.

http://www.pcgameshardware.com/aid,6..._overclocking/

[bowman]

Better be a one-off..

1.7V to reach 4GHz on the Idiot Edition, mediocre single-threaded performance..

If this is all true, at this point I'll end up waiting for Nehalem solely for the Core 2 price drop. Oh well, one can hope it's a combination of clueless OCers and early silicon..

[Extelleron]

Over at pc games hardware there was an article about overclocking Nehalem to 4 ghz with an extreme processor but it took 1.7 volts to reach that speed.

http://www.pcgameshardware.com/aid,6..._overclocking/

And they also say that it is stable on aircooling @ 1.7V.

Obviously something is wrong with that report. Not to mention that there is absolutely no way Nehalem will require an insane amount of voltage to hit 4GHz, while the latest Core 2s can hit 4GHz at 1.2-1.3V no problem. The core is pretty much the same, and with an unlocked multiplier the changed infrastructure should not be a bottleneck.

[JumpingJack]

And they also say that it is stable on aircooling @ 1.7V.

Obviously something is wrong with that report. Not to mention that there is absolutely no way Nehalem will require an insane amount of voltage to hit 4GHz, while the latest Core 2s can hit 4GHz at 1.2-1.3V no problem. The core is pretty much the same, and with an unlocked multiplier the changed infrastructure should not be a bottleneck.

Not so sure myself ... the IDF technical presentation revealed that Nehalem implemented all static CMOS for their circuit. Frankly, I am not well read on the difference between domino or static logic circuits... though I am studying up so I can hopefully bring some better explanation on the differences.

In a nutshell, Intel used static logic for the benefit of power consumption, taking advantage I suspect of the strong PMOS performance. When I have a better handle on the physics and the advantages, disadvantages I can speak more intelligently about the topic. Generally speaking, static CMOS circuits do not lend as well to clocking, from what I have been gathering so far.

Jack

[OBR]

My first impressions with own Nehalem ...

1. BIOSes and whole platform is still very immature ...
2. Only little raised reference clock causes massive unstability ...
3. Ocing of locked-multi Nehalems will be a problem ... Reference clock of Bus is not like a FSB - my max stable clock is 140MHz from 133MHz
4. If you want great clocking, you need a Extreme chip with unlocked multi
5. With these early BIOSes is Memory performance poor (or Everest dont know how to measure it?)
6. Power consuption is in idle with all power savings great, but in fully load is far away from 45nm Quads ...
7. Max stable ocing was on my chip about 3500MHz ...
8. Performance is various, read my previous posts in this thread ...
9. We have two months to launch, and i believe Intel will do many improvements to that date. Will see next month, with new BIOSes and maybe different mobo ...

[XS Janus]

Nice summary OBR. Thanks!
The best I've seen this far.

[bowman]

Yeah, I must say, that's a great summary and also lines up with the slightly cloudy leaks elsewhere.

[Macadamia]

And they also say that it is stable on aircooling @ 1.7V.

Obviously something is wrong with that report. Not to mention that there is absolutely no way Nehalem will require an insane amount of voltage to hit 4GHz, while the latest Core 2s can hit 4GHz at 1.2-1.3V no problem. The core is pretty much the same, and with an unlocked multiplier the changed infrastructure should not be a bottleneck.

1. Static logic leaks less, thus voltages don't seem that dramatic? This factor is a pretty big departure from Penryn.
2. Bloomfield like Barcelona is a big core. Big dies normally dissipate heat easier (provided distant hotspots and all) compared to Penryn/Wolfdale.
3. It's not as simple as that. Or OBR and co. would be getting 4Ghz OCs too.

[Donnie27]

1. Static logic leaks less, thus voltages don't seem that dramatic? This factor is a pretty big departure from Penryn.
2. Bloomfield like Barcelona is a big core. Big dies normally dissipate heat easier (provided distant hotspots and all) compared to Penryn/Wolfdale.
3. It's not as simple as that. Or OBR and co. would be getting 4Ghz OCs too.

Overclocking has NEVER been exact. Results can differ on processor from the same batch as seen with the Opteron 165's. My bud got 450MHz out of his 3500+, pull his out and stuck my 3500+ in his rig and only got 120MHz that was barely stable 90MHz stable.

You are right about the larger surface area.

[Macadamia]

Overclocking has NEVER been exact. Results can differ on processor from the same batch as seen with the Opteron 165's. My bud got 450MHz out of his 3500+, pull his out and stuck my 3500+ in his rig and only got 120MHz that was barely stable 90MHz stable.

You are right about the larger surface area.

I meant that the core became more complex, used more power and likely went in a different direction than clock scalability (at least for 45nm) compared to the Core 2 Duo.

Was comparing relative OCs, nevermind if they were consistent averages or not. I mean, E0s hit 4G on stock. So that's a pretty big shocking change they applied right there.

[Donnie27]

I meant that the core became more complex, used more power and likely went in a different direction than clock scalability (at least for 45nm) compared to the Core 2 Duo.

Was comparing relative OCs, nevermind if they were consistent averages or not. I mean, E0s hit 4G on stock. So that's a pretty big shocking change they applied right there.

QFT! I'm shocked as well. But there's no rush for me to buy anything yet. Even my C0 will do 4GHz with almost no more voltaged added and on a very cheap $89 motherboard, hehehehe! I'm running at stock because I like it quieter better. Mine is almost dead silent with no silent tricks at all.

[Macadamia]

Since Yorkfield is so smooth I wonder if Dunnington's gonna get more attention for a while.

Heck, they probably perform better with less (or equal) power usage.

[Donnie27]

Since Yorkfield is so smooth I wonder if Dunnington's gonna get more attention for a while.

Heck, they probably perform better with less (or equal) power usage.

Give any of the Intel systems a shot? See for yourself. Some of the "smoothness" or the lack there of can be attributed to user skillz.