SPCR

SPCR article: http://www.silentpcreview.com/Intel_core_i7_launched

Interesting Article Mike:) But as I am going for the really low power option, I can't help but being mostly excited about the SSD drive you received in the Intel-package

Could you guys do a test of the bundled HSF? Given how much larger it is than even the biggest of the Socket 775 HSF, I'm curious whether it actually performs in the same league as the less expensive aftermarket coolers. I'm sure the fan is still crap, but if you can run it at a low enough speed almost any fan is acceptably quiet for most people.

Silent computing is a tiny niche, and most manufacturers pay us no attention whatsoever. For the last few years, we've ridden on the wave of environmentalism - companies wanted to appear environmentally friendly, so they started creating lower-powered components. This worked brilliantly for silencers. Now, though, the companies are looking more at performance-per-watt. That doesn't work as well, because the companies make devices like the current high-end video card lineup, and these Core i7 processors, which are hugely fast but very difficult to manage in a silent system.

These processors are indeed very efficient if you spend lots of time running demanding applications, but what about idle and low-usage situations? I also don't particularly like quad-core processors. I do very little which is assisted by more than two cores; in truth I do very little which even uses two cores, but the extra responsiveness of a dual-core system is excellent. I look forward to seeing what the dual-core variants look like.

Tobias wrote:Interesting Article Mike:) But as I am going for the really low power option, I can't help but being mostly excited about the SSD drive you received in the Intel-package

This is obviously off-topic. I actually had one in my main system for about 4-6 weeks and was planning to write about it. But it's hard to suggest that it's better than the WD Velociraptor, which has 4 times the capacity and is 1/4 the cost.

Why?

OK, I have a suspended VR as a 2nd drive for data in an Antec P150. The SSD was cloned from the previous OS drive, another VR. Did the noise level improve? Maybe. I sometimes heard seek before, w/ the SSD, hardly ever, with only the data VR. Did it get faster? Maybe, maybe not. Boot time didn't really seem to improve -- I have lots of programs, and w/ either the SSD or the VR, it takes over 1 min. for Win XP to become fully operational. In general, the system is very speedy, w/ 4gb of ddr3, Intel E8500 (3.16ghz) and ATI3850 on an Asus P5E3 Premium mb, and virtually silent.

This i7 review package came with only one SSD, so the one in my system was pulled, its contents cloned to a VR again, so that the i7 and C2D system each had its own SSD.

Back to the 2 VRs in my system now, for a week... do I notice any difference? No. Considering that you can buy 4 VRs for the price of one Intel SSD, I can't really recommend it. I might try it in a laptop tho and see what it does for noise/battery run time, but for general use, the price of the Intel SSD is still way too high.

One thing -- the SSD drops power by maybe 5W.

I also was more interested in the X25-m SSD drive in that package than the Core i7 CPU

Can you do a review of just the SSD? I've heard some people complain of coil whine from them.

Also id love to hear more comparisons to a VR. Can you hard mount a VR in any way, including inside a SQD, and still have it be as quiet as the SSD?

Aris wrote:I also was more interested in the X25-m SSD drive in that package than the Core i7 CPU

Can you do a review of just the SSD? I've heard some people complain of coil whine from them.

Also id love to hear more comparisons to a VR. Can you hard mount a VR in any way, including inside a SQD, and still have it be as quiet as the SSD?

No noise at all from either of the 2 samples, and, no, a HDD with mechanically spinning platters at 10,000 rpm, hard mounted in a case, can never be as quiet as a bunch of IC chips! The point I make is that absolute silence, tho there it is in an Intel SSD, is not necessarily audibly better than a component that makes some noise. The main thing is to make it inaudible in real use.

From the article...

It's hard to tell exactly what is going on under the hood as CPU-Z does not allow each core to be monitored. There's no software that we know of which tells whether inactive cores in a multi-core CPU underclocks and undervolt or go to sleep.

I know RMClock has a seperate monitoring screen for each core. I don't know if it was designed to handle things like the i7 which actually has separate control for each core. But might be worth a shot. Too bad you can only look at one core at a time.

Some findings of the german hardtecs4u.com
Power consumption of VRM and CPU. Nehamlem is the new idle king. Waiting for efficient chipsets/boards in mainstream market now ...
Notice the higher power consumption at load with turbo mode on (86 W vs 114 W). This is due to the fact that the voltage was increased during load with turbo mode activated. They say it went from 1.15 V to 1.233 V. The Mike's screenshot shows 1.17 V, though.

jessekopelman wrote:Could you guys do a test of the bundled HSF? Given how much larger it is than even the biggest of the Socket 775 HSF, I'm curious whether it actually performs in the same league as the less expensive aftermarket coolers. I'm sure the fan is still crap, but if you can run it at a low enough speed almost any fan is acceptably quiet for most people.

I considered doing this but decided it against it for many reasons:
-There are too many different revisions. I count three LGA775 stock coolers in our lab that are all slightly different. Also I don't think any of them are for the quads.
-The i7 cooler can't be mounted on LGA775 systems and vice versa so we can't make a direct comparison.
-The Core i7's have a higher power draw so it's hard to make an indirect comparison.
-We only have one other cooler we can compare it to fairly - the TRUE-120. Not exactly fair...

I can say that the cooler it shipped with is significantly quieter than most stock coolers I've worked with. It is very quiet at about 7V and starts up somewhere in that region when plugged into the DX58SO with the 920 at idle. Usually the minimum speed is not low enough to be tolerable by SPCR standards, but this one surprising me. It is a bit clicky at 7V, but you can only hear it if you're within a foot or so.

Lawrence Lee wrote:

jessekopelman wrote:Could you guys do a test of the bundled HSF? Given how much larger it is than even the biggest of the Socket 775 HSF, I'm curious whether it actually performs in the same league as the less expensive aftermarket coolers. I'm sure the fan is still crap, but if you can run it at a low enough speed almost any fan is acceptably quiet for most people.

I considered doing this but decided it against it for many reasons:

I understand your reasons, but I think it would be nice to have as a benchmark going forward. You know a big question for any low cost heatsink will be: is this quieter than the boxed one I get for free. This won't be an issue until a more mainstream chipset comes out for i7, so you have plenty of time -- just something I'd like to see at some point.

What coolers besides TRUE (the version in the review, naturally) and Noctua NH-U12P (free mounting kit available from Noctua) are compatible with Core i7?

I don't mean all coolers, just the good ones

According to http://en.wikipedia.org/wiki/Intel_Nehalem#Variants the first SPCR-relevant dual-core Nehalem (Codename: Havendale) will be available in H2 2009. The power envelope of these are 75-95 W. Quite a lot of power for a CPU for most SPCR users. However, this is speculated to include an IGP on the CPU it self. So IMO, depending on the IGP, Nehalem/Havendale will be either a dud or great.

But for another year or so, Intel will have nothing better CPU-wise than the E7x00 and E8x00 series for most of us (at least for me). However, I think they're doing pretty well with their SSD line. I for one can't wait to see a more reasonable priced SSD with X25-M-like performance.

You also have to remember that 95w TDP includes the memory controller which was typically on the northbridge. Which is on par with current Core 2 Duo's if you add in the power dissipation of the memory controller currently residing on the northbridge.

If that 95w includes an IGP, that would make it VERY frugal as far as power consumption is concerned.

Aris wrote:You also have to remember that 95w TDP includes the memory controller which was typically on the northbridge. Which is on par with current Core 2 Duo's if you add in the power dissipation of the memory controller currently residing on the northbridge.

This is what people usually forget when they compare the TDP between AMD and Intel CPUs (Phenom / AthlonX2 vs Core2, mainly).

Aris wrote:You also have to remember that 95w TDP includes the memory controller which was typically on the northbridge. Which is on par with current Core 2 Duo's if you add in the power dissipation of the memory controller currently residing on the northbridge.

If that 95w includes an IGP, that would make it VERY frugal as far as power consumption is concerned.

Good point! I think you're right. I didn't think about MCH being moved to the CPU .. now if we could only have a review on a dual-core Nehalem like now.

so they made a prescott again with a slower FSB and more threads.

I am so thrilled.

for 130w and 8 threads with the ram hitting 12gb/s on my "old" prescott....
it may be time to upgrade in the next decade or so.

Mentioning extra consumption of physical ram with hyper threading should have been said.

A simple example: a 186mb file turns into 1.2GB if to float a search through the file as a variable in memory with hyper threading. That is the speed increase folks...

and for ****sake, don't ever call a pentium four a dud again. It will only make you dumber...by the timew they narrow down what worked...well, the i7 seems a good more factual direction then the core2 marketed success story. Note "marketed" success story...

The €i7 doesn't have a FSB.