Intel's Sandy Bridge-E To Ship Without Stock Cooler, Thermal Conclusions Questionable

Intel's Sandy Bridge-E To Ship Without Stock Cooler, Thermal Conclusions Questionable

Sandy Bridge processors have been available for eight months, but Intel's highest-end CPU remains based on the older 32nm Westmere architecture, rather than on SB silicon. A report from VR-Zone suggests Intel's Sandy Bridge-E CPUs may have been delayed for thermal reasons. According to the site, Sandy Bridge-E chips will ship without an Intel standard cooler and may dissipate as much as 180W. To date, Intel has said only that the upcoming processors will have a TDP in line with its previous high-end desktop chips.


The upcoming socket will span enthusiast and server systems.

Despite appearances, the two statements aren't necessarily linked and should be evaluated separately. Any decision Intel makes not to ship a stock cooler is going to be based on how often said coolers are actually used. The vast majority of boutique OEMs emphasize upselling customers on both CPU overclocks and CPU cooling, even in cases where the CPU is cooled more than adequately by the default Intel solution.

Intel's X79 chipset will use the company's all-new, high-end LGA2011 socket--a socket they'll share with upcoming Xeon processors that aren't expected to launch until early 2012. There's no indication that Intel won't have a stock heatsink+fan design ready for the Xeon products; the company's decision (if true) not to launch an enthusiast stock cooler in the face of widespread third-party support is almost certainly a business decision as opposed to a thermal one.



With that said, the question of CPU TDP deserves a bit of attention. When we reviewed MainGear's Shift last year, we measured the system's total power consumption running Prime95 at 4.2GHz (the speed of the system as shipped) and at 3.33GHz (the stock speed of the Westmere-based i7-980X.) In the Shift's case, raising the CPU clock by 26 percent increased total system power consumption by a whopping 37 percent. Much of that jump--about 60 percent of it--was driven by increased voltages. Given the rate at which power consumption can climb when voltages and frequency are both adjusted upwards, it's not unreasonable to think that top-end LGA2011 chips could draw comparably large amounts of power when overclocked. In its base configuration, the i7-980X is rated as a 130W TDP processor--a figure that seems reasonable, based on what we see above.

VR-Zone, however, states "these beasts are consuming closer to 180W and that's without even overclocking them. In fact, according to PSU design guidance we've seen, Intel is telling power supply makers to make sure their Sandy Bridge-E PSUs can cope with a peak current of 23A on the 12V2 rail..."

Being able to provide 23A of peak current across the 12V rail has precious little to do with a sustained 180W TDP. VR-Zone's figures, if accurate, imply that Intel's Sandy Bridge-E CPU's draw far more power than one would expect based on the performance of both Westmere and quad-core Sandy Bridge. It's certainly possible that the first generation of hexa-core Sandy Bridge chips missed Intel's thermal guidelines, but we're not willing to conclude that's the situation as of this writing.
0
+ -

I find it hard to believe that thermal issues would play a role in the absence of a OEM cooler.

These new chips throttle supposedly and with the die shrink said CPU's should run cooler if anything even taking overclocking into account.

I think it has more to do with cost saving as most users purchasing these high end CPU's have no need for the stock cooler and this way Intel can lower the price.

+1
+ -

ThunderBird:
and this way Intel can lower the price

Nope,...not within their understanding when it come to Extreme CPU's.

I can hear it now,......"One Thousand Dollars A Pop, Please!"

0
+ -

180W TDP yikes! I thought they were supposed to be way more energy efficient than last gen, not sure why I thought that. I really don't care if they don't ship with a stock cooler, the intel stock coolers have all been terrible anyway. If you can afford one of these chips you can afford to spend a few extra dollars on a decent cooler.

I agree with @realneil the price won't drop because of the cooler. If the chips are powerful they will charge premium for them.

0
+ -

I've got to disagree with people who claim Intel stock coolers are "terrible." I hate the LGA push-pin mount method with an absolute passion, but that's a mounting design flaw, not a problem with the cooler's performance. I haven't had a performance issue with a stock cooler since Prescott--the Core 2 Duo coolers and Core i5/i7 models that followed them have all been more than adequate.

Sure, it's possible to get better performance with a relatively cheap ($25-30) aftermarket cooler, but boxed retail CPUs almost never cost more than OEM counterparts. It's a win/win situation.

0
+ -

Joel H:
I've got to disagree with people who claim Intel stock coolers are "terrible."

I don't think that. I just know from past experience that they are sometimes not good enough when you push the CPU beyond it's factory settings. They certainly do the job as designed, at stock speeds.

Personally, I like the Corsair H50's and H60's. I put two high output/low noise fans on them (push/pull) and they do the job on Overclocked CPU's, quietly. It costs more to do it, but it's worth it to do.

0
+ -

Real:

http://hothardware.com/Reviews/Digital-Storms-Core-i5-System-Reviewed/?page=13

Check pages 13 & 14 for info on push/pull performance vs. single-fan usage. Push/pull can have a thermal impact, but depending on the fans you're using, it may not--or the added noise may not be worth the temperature reduction. Using two fans to move air across a radiator creates far more wind resistance than just using one--so much, in fact, that you may see higher temps from doing so.

(This depends on the speed and shape of your fans. YMMV.)

0
+ -

Joel H:
Push/pull can have a thermal impact, but depending on the fans you're using, it may not--or the added noise may not be worth the temperature reduction.

You're right, as I've learned through trial and error. I use High Airflow/Low Noise fans to do the job for me.   These Fans   They're 18 dBA Quiet, yet they move 69.15 CFM of air at speed. I have them exhausting air, rather than pulling it into the case. They run on full power all of the time, but they're reasonably quiet.

0
+ -

Yea Intel will not budge on the Extreme series CPU $999 all day.

I was thinking if they came out with a non Extreme series perhaps but who knows.

0
+ -

"I don't believe this report. How could a new Processor thats 10nm less in size be more power hungry than for example , the current 32nm 990X, Both are 6 Cores. And why would Intel, Not include a cooler, that would be stupid. MY argument also stems from Sandy Bridge K chip being so efficient yet they don't run that hot at high clocks (4.5 -4.8). The extreme cooler included with the 990x is a great one, why would it not be more than sufficient for SB-e."

-Optimus

0
+ -

Intel can keep their metal and plastic if it means that the price is $20 less... Of course this is Intel so that is not an option.

0
+ -

Even though the Xtreme Edition cooler is pretty decent and it should be included in a $900+ CPU .Hard to imagine why it would not be sufficient to keep the temp cool enough.with SB-E.They can actually keep the stock generic push-pin joke all together.Do no think that by Not Including a cooler Intel will pass the savings on..

Looks to me like they missed the guidelines at this time.Plenty of time for Intel to sort it out and the coolers will not be missed or noticed as much as noticeably higher thermals and power draw.

What is most likely to be 'questionable' is the 30-60 day grace period for those that have to get the new chip right away

Without a doubt I know I can count on the HH team to keep us updated and on the Inside track..with the upcoming developments.

+1
+ -

Optimus,

Have you noticed CPU clockspeeds increasing much over the past five years? Intel's Core 2 Duo Conroe hit 3GHz. These upcoming $999 chips are looking at base clockspeeds of 3.4 - 3.5GHz. Conroe was a 65nm chip; Sandy Bridge's process is only 1/4 the size.

The short answer is this: As we've added cores and shrunk process technology, CPUs scale more poorly as voltage is increased. This is due partly to what's called gate leakage, and caused partly by the fact that adding a millivolt means a lot more at 1.2v than it did in the days when CPUs ran at 1.8v. Furthermore, raising core power consumption by 5W per core means a total increase of 30W if all six cores are operating.

I don't think the report is accurate, either, but your response indicates you aren't aware of prevailing trends or technological constraints. The whole reason Intel opted to start branching into higher core counts and away from clockspeed is because it's become easier to add CPUs to a die than it is to keep ratcheting up the clock.

0
+ -

I think that they are asking for more power on the 12V rail because they will have more aggressive turbo clocks. A max CPU usage of 180W for several seconds might cause the PSU with smaller rails to automatically shut down, but will not severely impact the CPU.

Login or Register to Comment
Post a Comment
Username:   Password: