Intel's LGA1156 and Lynnfield core

[smilingcrow]

“Turbo Boost was disabled during testing — we consider it overclocking.â€

[Meato]

I hope to see you guys test the Gigabyte and MSi uATX s1156 motherboards. Would love to see how they stack up voltage-, stability-, and performance-wise.

I second that request.

[PartEleven]

Clock-for-clock comparisons usually aren't very interesting, since architectural differences mean you'll get different real-world performance per-clock from different CPUs. Ofcourse, Bloomfield and Lynnfield have pretty much the same architecture, but the biggest difference between them is arguably the much improved turbo mode.

I guess to me, that is the entire point of CPU reviews. If SPCR did comparisons with Turbo mode enabled, then how can I distinguish whether its performance advantage over Phenom II (didn't bother to verify this; just take it as a hypothetical example) is due to architecture improvements or purely clock speed? It could very well be that the performance improvement is purely clock speed, and that clock-for-clock it's actually slower than Phenom II (again, just making hypothetical arguments here). The in that case I could actually get a better value by buying the cheaper AMD cpu and overclocking it to the i5's Turbo mode speeds. But I would have no way of knowing that, since the review compares a higher clocked i5 to a lower clocked Phenom II. Yes, I suppose I could try to track down some other review that has a benchmark number with Phenom II was tested at the same clock speed as the i5 with Turbo mode on, but's a lot of extra work.

I don't know about you, but I am getting a bit tired of the "real-world" arguments against testing methods. A good method should enable you to extrapolate the results to a variety of different situations. While a "real-world" test would only let you make comparisons under the same exact conditions.

Apparently you can't satisfy everyone unless you test every conceivable combination. One of Anandtech's writers had to make a blog post to address the clock-for-clock issue. Just as there are people here complaining against disabling Turbo mode here, there are users complaining about enabling it on Anandtech.

EDIT: Here's a quote from Anandtech that sums up pretty well how I feel about this

Anyway. Personally, the test combo that I'm more interested in is testing with Turbo off as well as on. Since the performance boost provided by Turbo is variable based on whatever other apps the user or operating system may be multitasking in the background, testing with Turbo off provides the "guaranteed minimum", i.e. a baseline level of performance that a Turbo capable CPU will provide, and that any Turbo boosted extra is considered a free bonus.

Haha, this post is getting into the tl;dr range.

[Mats]

A great review as usual!

How hard can it be to make it right tho? Run the tests twice, with Turbo on, and then off. No one can complain about that.

[PartEleven]

Another thought just occured to me. There's plenty of other quality sites reviewing this chip, do we really need SPCR to do a comprehensive test of its abilities?

[smilingcrow]

Here's a quote from Anandtech that sums up pretty well how I feel about this

Anyway. Personally, the test combo that I'm more interested in is testing with Turbo off as well as on. Since the performance boost provided by Turbo is variable based on whatever other apps the user or operating system may be multitasking in the background, testing with Turbo off provides the "guaranteed minimum", i.e. a baseline level of performance that a Turbo capable CPU will provide, and that any Turbo boosted extra is considered a free bonus.

And the more sites that review in both modes the more data we have to see how TB performs in practice. If it’s not tested thoroughly how can we determine if it lives up to what appears on paper to be a promising technology?

[smilingcrow]

Another thought just occured to me. There's plenty of other quality sites reviewing this chip, do we really need SPCR to do a comprehensive test of its abilities?

I look at the SPCR review mainly for the more detailed power data ( 2 and 4 cores loaded etc) and under-volting info and not for performance data. But because it was only reviewed with TB in one state I don’t know how TB affects these parameters. I guess that with TB on you can’t under-volt so much as the higher clock speeds will likely make a difference. It would be good to see what negative impact TB has as well as the positive.

[PartEleven]

Well, just because it is good to have thorough testing, doesn't require EVERY site to comprehesively test every ability of the chip. This particular cpu had so much press coverage going into its launch that I'm sure there is more than enough coverage for you to do cross-site performance comparisons.

You make a good point about the detailed power consumption data though, since SPCR is one of the few sites that specializes in this.

[yacoub]

Check this out: http://www.anandtech.com/weblog/showpost.aspx?i=642
Note also Gary's comments beneath the article to get an idea of what they have coming up (P55 motherboard roundups, possibly a P55 onboard audio chipset review later this month.).

Between SPCR and Anandtech, all the info we could want should be available shortly.

[mentawl]

The multiplier is locked on the current Lynnfield CPUs so turbo mode can only work by increasing the cpu base frequency which you could call clock speed.

Actually, you're wrong there sir. Turbo boost *only* increases the multiplier, it doesn't touch the base frequency.

[bradc]

I agree, I got to Anand and Techreport for performance and platform info, then come to SPCR for power consumption, idle states etc.

I too beleive that sites should show figures with turbo boost enabled and disabled, it can make things a bit hard to compare otherwise.

Personally I think turbo boost is great, remember when P4's were at 3.6 and 3.8ghz (yeah they were crap but whatever) and PD's were only at 3.2ghz. At the time going to an extra core cost you quite a few mhz.

Then with C2D's and C2Q's we are now able to get 3.33ghz Duals and only 3ghz Quads (ignoring the 3.2ghz EE)

Even ignore the highest cpu models, at the same price point the choice is similar at the moment, to get extra cores you lose raw mhz speed. IE the E8500 at 3.17ghz is the same price as the Q9400 at only 2.66ghz.

For the first time with this new platform it is possible to get a quick quad core cpu (and with HT too in the i7 8xx series) that is capable of parallel tasks, and is capable of speeding itself up when you are only using a few cores, and becomes equivelant of a fast dual core. IMHO this is perfect.

[frostedflakes]

Another thought just occured to me. There's plenty of other quality sites reviewing this chip, do we really need SPCR to do a comprehensive test of its abilities?

Exactly. No offense to MikeC and the staff (and I'm sure none will be taken, because they realize this), but SPCR isn't exactly well known for their performance benchmarking. Obviously the focus for a site like this tends to be on noise, heat, and power consumption, and this comes at the expense of overclocking and performance testing. Benchmarks with Turbo Mode enabled would be nice, but this is something that's covered in much greater detail by pretty much every hardware site out there, so it's hardly necessary. And the people claiming that Turbo Mode testing was omitted because of bias or something like that... seriously? Come on guys. They were just attempting to provide an apples to apples performance comparison between 1156 and other platforms.

[aristide1]

Ugh, it looks like every review site got sent an i5-750 and an i7-870, but I can't find any benchmarks comparing the i7-920 to the i7-860, which share a price point. Obviously, the 860 will be the winner in power consumption, but I still want to see the performance numbers.

There should help, somewhat:
http://www.xbitlabs.com/articles/cpu/di ... 7-870.html
So trim a few % off the 870 numbers, how far off could you be?

HardOCP included an i7-920 in their review. Bottom line - at the end of the review they point out Intel no longer shows the i7-920 in the product lineup going forward because it's redundant against the i7-860.

That's not good. The P55 chipset has 16 PCIE lanes available, so you can have 2 slots with 8 lanes each. The 1366 board? You can get quad slot boards. Even the X38 chipset for 775 has 32 lanes. This is skimpy for those who want folding farms. Too bad the 920 eats more watts than the 860.

[loimlo]

Guys, Etiquette is the world language I believe.

Frankly speaking, I don't think Turbo mode is worth arguing for possible performance loss -- 1156 already beat 775 and AM3 in this review without Turbo. You could estimate on your own that the advantage should be greater when enabling Turbo mode. And outcome will still favor 1156 regardless of Turbo on or off. There's only relative poor or poorer comparison result for 775/AM3. That said, I hope SPCR would consider improving power consumption test which shall include Turbo enabled in the future reviews. This would make a great comparison for silencers who bears power figures in mind.

Also, Motherboard Impressions on last page is very useful, informative for silercers like me! Thank you for taking advice about exotic speedstep setting! Besides, all the nice niches such as PWM fan, Speedfan software, Turbo mode are very helpful to a potential buyer! These small but useful functions are scarce if not none to find across the reviews and forum discussions. To make matters worse, these features are board by board specific. Though most people never pay attention to such subtle differences, these features often make a good board a great one. Driven by OVERCLOCKING trend, most vendors/reviews tend to lack of these details and concentrate on extreme OC. Sigh. Last but not least, I anticipate SPCR to inspect phase changing mechanism across motherboard vendors. I'm sure MSI is the best, but I would love to hear SPCR's approval.

I DO appreciate your involvement in this article, especially subtle things.

[~El~Jefe~]

I agree, I got to Anand and Techreport for performance and platform info, then come to SPCR for power consumption, idle states etc.

I too beleive that sites should show figures with turbo boost enabled and disabled, it can make things a bit hard to compare otherwise.

Personally I think turbo boost is great, remember when P4's were at 3.6 and 3.8ghz (yeah they were crap but whatever) and PD's were only at 3.2ghz. At the time going to an extra core cost you quite a few mhz.

Then with C2D's and C2Q's we are now able to get 3.33ghz Duals and only 3ghz Quads (ignoring the 3.2ghz EE)

Even ignore the highest cpu models, at the same price point the choice is similar at the moment, to get extra cores you lose raw mhz speed. IE the E8500 at 3.17ghz is the same price as the Q9400 at only 2.66ghz.

For the first time with this new platform it is possible to get a quick quad core cpu (and with HT too in the i7 8xx series) that is capable of parallel tasks, and is capable of speeding itself up when you are only using a few cores, and becomes equivelant of a fast dual core. IMHO this is perfect.

This is my argument for a long time. Quad is near nothing in performance increase compared to a much cheaper e8500. I know for a fact that 99.8% of the time a quad is near useless for much. This lynfield looks promising. I wouldnt get it without the hyperthreading though. That really is a waste of cash to upgrade without going for the new features. It is like buying a 6700 as your main rig, no VT, no extra cache and all for a dismally small price difference than a e8400.

What was not shown was the 285 dollar i7 Lynnfield. that is proper upgrade! If someone has 775 and wants to get a quad for business purposes, it is not throwing money away. Same board, thats 150 dollars saved, same ram, well, thats like 50 dollars saved. And no reinstalling, thats 2 days of lost work or life at times. The $285 dollar chip is the one that will make Intel the most cash. That's a sweet deal.

(btw, having no IDE and no floppy is really lame. That could have cost a company about .75 cents more to put on.)

[bradc]

Well for me, I have an i7 975 in my gaming system, an E8400 in my everday use system and another E8400 in my server. I'd like to update the motherboard in my everyday system from my now ancient P5N32-E SLI motherboard, but the problem is that I need either 2x SAS ports or 3x PCIe slots (8x lanes for each), so Lynnfield still doesn't do it for me for that computer system, I might end up getting an i7 920 with a mid range motherboard for my desktop.

My server spends a bit of time doing a specific task that only uses one core, so Lynnfield with it's super turbo mode might be very handy for that, but again I need 3x PCIe slots for the pair of raid cards and graphics and I'm not sure about using the slower southbridge lanes. Not for their bandwidth, but for the bandwidth from the CPU to the SB.

Looks like 1366 for me all round

[AZBrandon]

That's not good. The P55 chipset has 16 PCIE lanes available, so you can have 2 slots with 8 lanes each. The 1366 board? You can get quad slot boards. Even the X38 chipset for 775 has 32 lanes. This is skimpy for those who want folding farms. Too bad the 920 eats more watts than the 860.

I don't think they really care about that particular market, to be honest. P55 still supports dual GPU's at 8 lanes each, and I know HardOCP in particular is working on a review right now to benchmark a P55 i7 against an X58 i7 each with dual GPU's so the only difference is 8 lanes versus 16. That review should tell us if it really makes a difference. I suspect, as it sounded like their reviewer suspected, that very few applications will make any difference in having the extra bandwidth.

So from a marketing standpoint it means you can run dual GPU on the P55 platform, possibly even 3 video cards if one runs at 4x (which we know is sufficient for FAH) and for anything more than that you'd just have to bit the bullet and go X58 with either a leftover 920 for however long they keep selling them or step up to the 950. If you think about it, from a marketing standpoint, anybody wanting to run 4 video cards probably has enough money to do an i7-950 anyway, or at least so goes the logic.

Maybe it's Intel's way of giving AMD a teeny, tiny market niche - cheap mobos and cheap CPU's that can run 4 PCIe slots.

[aristide1]

So from a marketing standpoint it means you can run dual GPU on the P55 platform, possibly even 3 video cards if one runs at 4x (which we know is sufficient for FAH) and for anything more than that you'd just have to bit the bullet and go X58 with either a leftover 920 for however long they keep selling them or step up to the 950. If you think about it, from a marketing standpoint, anybody wanting to run 4 video cards probably has enough money to do an i7-950 anyway, or at least so goes the logic.

Well the really serious P55 entries already have more than 16 lanes with the extra nForce 200 chip handling the additional lanes. The ironic thing is these beasts start at $200, same place the socket 1366 boards start with that have 4 PCIE slots with 8 lanes each.

http://www.hardware.fr/medias/screensho ... IMG0026749

Also - I hate it when companies have huge successes with what boils down to a product that's been dumbed down.

By the way if you get a nice top view of some of the new P55 motherboards you may see 2 sets of holes for mounting the cpu cooler. The clever ones will accept the new coolers or the socket 775 coolers.

[yacoub]

Here are two update articles from Anandtech.
The first one mentions overclocking and power consumption data.
In the second one there's a nice comprehensive run of charts showing of where the three new Lynnfield CPUs slot in performance against a nice array of other CPUs:

http://www.anandtech.com/cpuchipsets/sh ... spx?i=3640
http://www.anandtech.com/cpuchipsets/sh ... spx?i=3641

[RBBOT]

Whether or not turbo mode should be defined as "overclocking" for whatever precise definition of the word you want to use is irrelevant and shouldn't determine if it is enabled during testing. Also, a comparison of the clock-for-clock performance of CPUs is fairly uninteresting, and should not really be a goal of the review.

What I'm looking for from a review of a CPU on this particular site is a review of the product in the configuration that gives the most optimal performance and good reliability whilst producing what the expert reviewers here feel is an acceptable amount of heat to still cool it quietly. Whether it is defined as "overclocking" or not isn't important, only whether it produces too much heat gain. If turning it on was found to increase the heat produced significantly, then that is the sort of thing I, and probably most of the readers here would like to know about. Hence I was disappointed that it was switched off before the testing even began.

If possible, I'd like to see a follow-up review on Windows 7 where the OS tries to actively keep physical cores idle (ideally 64-bit as the world is moving in that direction) that explores the effect of enabling/disabling turbo mode on the thermal performance and work done per watt, as SPCR is going to do a much better job of that comparison than most other review sites.



Also, PCPro have published a first review of a Core i7 laptop range and the effect of turbo mode here is far more significant than on the desktop processors. There are three chips in the range clocked at 1.6, 1.73 and 2Ghz. The respective turbo modes are 2.8, 3.06 and 3.2Ghz so that is a 60%-75% increase.

http://www.pcpro.co.uk/blogs/2009/09/18 ... rst-review

[smilingcrow]

Also, PCPro have published a first review of a Core i7 laptop range and the effect of turbo mode here is far more significant than on the desktop processors. There are three chips in the range clocked at 1.6, 1.73 and 2Ghz. The respective turbo modes are 2.8, 3.06 and 3.2Ghz so that is a 60%-75% increase. http://www.pcpro.co.uk/blogs/2009/09/18 ... rst-review

Without Turbo Boost those 45nm mobile quads wouldn’t make much sense as their performance with applications that only utilise 1 or 2 threads would be farcical given the pricing. The entry level 1.6GHz version is ‘only’ a 45W TDP versus I think 55W for the other two. Dell UK have a good deal on a Studio 15 with the 1.6 version and a decent spec for about £650.

[colm]

I stopped at page one.

prescott inefficent? its like swapping a car because the bad battery couldn't hold a charge. uh oh, radiators leaking. swap the car, put the same bad stuff back together again...blame anything but the body holding the parts...

these sockets since 775, its like a slot one only more stubborn to fade into the realms of truth: complete failure in history proven. Inefficient is an understatement for lack of actual non-bleeding connection.
You can try and challenge my 11 years and 100k pc hours. it is futile..


I am off to 30000 hours with a new 478 board. oh my god, it has all the vrms this time. If it ran with only 75% of its volts at 6.5 gbit memory bandwidth for five years on less than 400watts...what it is going to do with actual correct power and a case painted by someone who knows them...

I may drop back to 250 watts for the full atx, and 725mhz vid card...3400 mhz at, oh no, the big bad scary pinned 103 watt cpu...
the precott is the dual core, the precott has the 45nm, the precott also matches mobos with 90nm..all in the same cpu. Read facts before making falsed market statements.

I am vented, thanks for the review. And oh, there is an article confessing at toms hardware about an 1156 error and 600 dollar pos called a cpu ...they pretended the problem is solved. whip out those walletts "evolutionaries" whip it out and spend badly...crazier than a slot 1 you silly fools.

[smilingcrow]

I stopped at page one.

prescott inefficent? its like swapping a car because the bad battery couldn't hold a charge. uh oh, radiators leaking. swap the car, put the same bad stuff back together again...blame anything but the body holding the parts...

.........................................etc

I'll have what he's been drinking.

[cmthomson]

I stopped at page one.

prescott inefficent? its like swapping a car because the bad battery couldn't hold a charge. uh oh, radiators leaking. swap the car, put the same bad stuff back together again...blame anything but the body holding the parts...

.........................................etc

I'll have what he's been drinking.

Or smoking.