Intel's 'Overspeed protection' will limit Core i7 overclocks!!!!!!

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Heres the whole proble. Every bench weve seen is done on a engineering sample. Are they the same as retail? I believe you can change the bclk, but that may be it, for the lessor models anyways. Guess well still have to wait for sure, but it still doesnt look good
 
Not being an electrical engineer, but seems as though the throttling is done entirely on the CPU side, motherboard has nothing to do with it. CPU measures voltage and amperage, calculates whether it's using too much, and scales back, thus eliminating overclocks past 130 watts (Talking non-EE on all of this.).

I don't think you can get around this, roadrunner. But I'll be happy to see if they can/do.
 
Good news!! "I am to be VERY VERY clear, there is not Overspeed Protection, we have a BIOS setting to bypass the PCU control of Current and Power. On smackover, it is call "CPU VR Current Limit Override", and it will allow you to bypass the thermal and courant limites on the Core i7 920 ans 940. Who ever claim the otherway around did not read the specifications enough hehehhe
we have recommanded to the motherboard makers to follow our steps.
So, relax, and don't worry ...

Francois"

So the bios option IS doable
 


Are you saying that you can enable "CPU VR current override" on all three, the i920, i940, and i965 and none are crippled, or are you saying that only the extreme i965 has the CPU VR current override and that the 920 and 940 are crippled?

EDIT: LOL...never mind per post above mine...had I just waited 5 more mins to post.
 
Yes, there is no restriction on any of the CPUs.
A few sites were hitting 4.0Ghz on the 920.
They also did not mention the limit mentioned in THG.

I think the big issue at the moment is the limited number of mobos available for testing.

Some likely have this feature.
Some do not.

Some OC better than others.

I think it will take a few weeks to really get a good idea of the final results.

What is clear is that at launch the i7 will not be the choice for those that are price sensitive. For those who can afford a premium, will have a great chip.

 


Me too and I'm an AMD fan.

I can't wait for Deneb, but even when it comes out, I'm quite sure that Nehalem will beat it. That's because the die shrink and optimizations won't be a new architecture. If it gets a 20% boost over B3, then it will match the best of Core 2 but fall behind Nehalem.

Nehalem will compete on the server space and AMD will be in trouble. It will be awhile before i7 trickles down to the mainstream desktop, and that's where Deneb will have to survive.

What AMD has going for it are inexpensive but great chipsets and the best GPU's around. June 2008 should bring out new GPU's that will continue the trend. Deneb should be decent enough for budget builds, but it will be competing against Core 2, not Nehalem.

At any rate, a Black Edition Deneb won't hurt Nehalem all that much. Without an unlocked multiplier, some sites are reporting a stable 3.8 over 2.66, and with the unlocked multiplier of the EE, 4.5 over 3.2. I doubt that a 3.0 Deneb will match that.

Thunderman, I'm as big an AMD fan as you are, but I'm a realist. I once read that sanity is the unconditional acceptance of reality. Accept the reality that AMD's Phenom's are improvements over the Athlon X2. They are not Core 2 killers (but match Core 2 in some benchmarks and markets). They will never be Nehalem killers.

When AMD's next architecture arrives by 2010 and competes against either Nehalem's tock, or Intel's next tick; then we'll see how the companies stand in having the best CPU architecture on the planet.

Until then, we know Intel has the best architecture.



Does a dual GPU card count as hard core gaming? Maybe i7 will help propel dual GPU cards to the mainstream, as ATI brings out more models (4870x2 then 4850x2 followed by new cards in June 2009).

I hope Deneb's as good as Mu_Engineer says, but I still don't think it will beat Nehalem. That said, beating is relative if Nehalem platforms are much more expensive than Core 2, which are slightly more expensive than AMD.

Tom's review made my 8750 look sad, whereas it looked quite good compared to the Intel CPU's it was targeted towards at the time. Still, in games with a single GPU card, i7 doesn't look to be the price performance winner.
 


All benchmarks are conducive towards Intel because they currently have the better processors. Just in the same way the Hazel Park Michigan Harness Raceway is conductive to horses over mules; I'll give you a hint, it's not the dirt or the shape of the track.
 
For that analogy to work TC, AMD CPU's would have to be hybrids of a horse and a donkey. AMD didn't buy Via, so that's not actually viable.

Let's compare horses to horses here. AMD CPU's are plowhorses, good enough for everyday work, and Intel CPU's are thoroughbreds good for winning races.

If we were comparing cats to cats, then it might be lion vs. tiger. When comparing AMD to Intel using animal analogies, choose carefully. The days of fruit analogies are over, but there's still much work to be done getting animal analogies to work. :pt1cable:
 
Are you sure non-extreme versions' overlimit protection can be overridden? Tom's hardware used a board with the feature, and had it turned on, it just didn't work for non-extreme versions.
http://www.tomshardware.com/gallery/bios_no_130_a_130_w,0101-165532-0-2-3-1-png-.html
http://www.tomshardware.com/reviews/Intel-Core-i7-Nehalem,2057-7.html
All the other sites used the "extreme" 965 chip for oc test with the override, and didn't try it on lower models.

Tom's is saying it works with "extreme" version, not with others. Other sites gave no information. The quote from that one Intel employee is also unclear, as it didn't specify if it's referring to "extreme" or non-extreme version. But if it's referring to all i7, that would mean Tom's had the actual cpus, tried it with overlimit protection overridden, and then decided to tell a white lie. It's just unlikely.
 
Considering intel telling us there not limited, which makes sense.
From what i see its a thermal limit that does not effect OC until you hit the point where youd fry your CPU.
It looks like intel is trying to cut down on people trying to get them to exchange a burnt OC'd CPU maybe by overclocking 2 far.

Anyhow if this is how it is, i plan on 940. If not ill go extreme. Just wish they'd release them now instead of 17th. Seeing some people get them and not being able to order online drives me nuts.
 


Actually no he isn't. From other reports earlier on the Overspeed Protection has been disabled for all Core i7s released. It MAY be implimented later but I doubt it unless they are low end Celeron like models.



http://www.tweaktown.com/reviews/1642/14/intel_core_i7_nehalem_arrives_and_fsb_departs/index.html

Look at that. In Crysis on very high @1600x1200 with a 9800GX2 and a Ci7 920 @ 2.66GHz that keeps up with a QX9770 @ 3.2GHz. The i965XE hets 10.11FPS better which is a 35%, roughly, performance increase. And considering the i920 is only about $300 vs the QX9770 @ $1500 thats one hell of a price/performance ratios even with a 540MHz disadvanatage.



No its with the X58 chipset so far. I have yet to see either chips being limited in OCing. Show me a i920 thats limited and I will agree. But until then there is no limiting.
 


By 2010 it will be a tock, not Nehalems Tick. The Tick is Westmere (32nm shrink of Nehalem thats said to be well on its way). The Tock in 2010 is Sandy Bridge thats supposed to be 16 cores 32 threads.
 


Per die probably...
 


You missed Anandtech perhaps? That's where I got my info on the 2.66 Nehalem's clocking to 3.8:

Our second overclock test utilized the 920 processor that clocks in at a stock 2.66GHz. The multiplier is locked on this CPU and the 940 model so overclocking is done via Bclk. We were able to reach a 24/7 stable 3.8GHz overclock on 1.5V with memory (BIOS 0503 raised our voltage requirements) at DDR3-1520 (7-7-7-20) on 1.675V. We think 4GHz is possible on this board with additional tuning and a BIOS update. However, Bclk is limited to around 200~220 on the current i7 series, so additional headroom is probably limited on this CPU. However, performance was excellent during overclocking and a 3.6GHz overclock was possible with 1.425V. This CPU reminds us of the Q6600 at launch, an excellent overclocker that continues to be a bargain.
Click to expand...

http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?

It will be a bargain once motherboard and DDR3 prices drop.

My hope is that Deneb at 3.0 stock will overclock like this particular Nehalem, to 3.6 or 3.8. If Deneb's IPC matches Core 2, then it still won't beat Nehalem's EE, which establishes the crown for top CPU, but if it overclocks well, then it will be a bargain, considering the cost of AMD motherboards.

Heck, we won't know till Deneb's out. Then we can make chimp vs. gorilla comparisons. (Just say no to fruit analogies. Leave apples to apples and apples to oranges to the vegans).
 
Tomshardware just put up an article stating that the overspeed protection can in fact be disabled. This isn't going to be a problem.
 


So they took it back. Nice.

Intel has added an Overspeed Protection to its Core i7 processors, keeping them from exceeding 130W or 100A. The company says it’s a safeguard in environments where that much power or current could result in a problem. The workaround is a BIOS switch. Overclockers will want to disable it before trying their hand at faster frequencies.

Our Core i7 920 sample is stable at 3.8 GHz and with a bit more tweaking would likely stabilize at 4 GHz as well. We didn’t really have any better luck with the 965 Extreme, which could be tweaked through its multiplier settings versus the Bclk-only, like the 920. But then again, this brief editorial was more a personal effort to address some of the comments brought up in our initial launch story.
Click to expand...

Just as I stated. The Overspeed Protection is mainly for enviroments where OCing will not occur to keep the Tirbo Mode from going overkill and frying the CPU. It can be turned off via BIOS.

Whats nice is to see a 2.66GHz CPU @3.8GHz stable on air. Looks like the Core i7 i920 might just be the Nehealem version of the Q6600.

Oh and thunderman (and dagger) looks like you jumped the gun a bit too early. Thats what happens. need to wait for all the details first before making assumptions.
 
TC, you know what I meant. I dont have the numbers, but Im betting the P4 beat out the K8 in super pi, and thats what I mean. Just like 3DMark. Different Arch's react to different apps differently, especially one so single purposed as super pi. Like has been said, using gaming to determine a cpus performance is similar to super pi, as Phenom beats i7 in a few games, in certain setups
 



Sir - I believe you need to reexamine the issue. The most thorough review I've seen yet: http://www.legionhardware.com/document.php?id=786 indicates this is GPU driven rather than a function of the processor. Drop it down to 8x6 resolution, and the new proccy can *still* put the bottleneck on a GTX 280.

So if your position is "Well, the new processor can't actually deliver any improvement on a system level because the Graphics Cards... So there's not much reason for a gamer to buy one until nVidia/ATI can catch up...", then you have an arguable case. But if you are taking the position the limit has to do with the architechture of the processor itelf, then you may do well if you took some time to educate yourself further.

As far as the rest, Deneb had better be a *huge* improvement. AMD's biggest strength has been in the enterprise. But looking at the results I've seen so far, the difference between Green's and Blue's offerings has become rather... stark.
 
I find it humorous that thunderboy used such an Intel-biased article to try to make AMD look good.

For example, the title of the article
Nehalem - Intel's new Core i7 gives them a big new lead
Click to expand...

Then it begins with this paragraph which is definitely Intel-biased. 25% increase clock-per-clock over Core 2? For single threaded applications you'll be lucky to see a 10% increase unless there's intense memory usage. 2.5x better than Phenom? I'm looking forward to Nehalem but these numbers sound pretty crazy to me.
Santa Clara (CA) - Today, Intel took the embargo off its latest Core architecture revision, Nehalem (pronounced Nuh hay lem). The reviews are in and Core i7 is a force to be reckoned with. Perhaps Bert Toepelt of Tom's Hardware sums it up best, "Put bluntly, you'd need two and a half Phenom X4 processors to compete with Intel's current Core i7 flagship model." With hyperthreading's eight virtual cores, a 25% increase clock-per-clock over Core 2, a new on-die memory controller and lower power consumption, does it have a downside? There are no pin compatible models to Core 2, meaning it requires a new motherboard (which currently costs more than the CPU). And, it only works with DDR3 memory.
Click to expand...


 
I dont have the numbers, but Im betting the P4 beat out the K8 in super pi
Click to expand...
As long as we're betting nickels in the context of 1M... you lost!

http://www.pcstats.com/articleview.cfm?articleid=1790&page=8. Second chart on the page - 3.2GHz Prescott (1 MB L2) loses to 2.4GHz Venice (512K L2).

On this benchmark forum - http://www.devhardware.com/forums/benchmarking-tools-23/hwbot-superpi-1m-ranking-93516.html - we can roughly gauge the clock efficiency of various architectures at Pi calculation by multiplying MHz by seconds to yield "million clock cycles to calculate 1M" (lower is better efficiency). I added various results from other sources (e.g. http://www.devhardware.com/forums/benchmarking-tools-23/devhardware-super-pi-42201.html) to round out the timelines, as this benchmark is very easy to replicate, though different versions and lower level memory settings can slightly affect times:

Athlon Thunderbird @ 1400 MHz: 104s (145600 MHz-s)
AthlonXP Palomino @ 1400 MHz: 79s (110600 MHz-s)
AthlonXP Barton @ 2656 MHz: 38.2s (101500 MHz-s)
Athlon64 San Diego @ 3180 MHz: 27.1s (86200 MHz-s)
Athlon64x2 Toledo @ 2864 MHz: 29.9s (85600 MHz-s)
Athlon64x2 Brisbane @ 3343 MHz: 26.3s (87900 MHz-s)
Phenom 9950 @ 3905 MHz: 19.9s (77700 MHz-s)
[Phenom WR result from http://www.xtremesystems.org/forums/showthread.php?t=206289]
Deneb @ 2801-3441 MHz: 25.1-20.5s (70-71000 MHz-s) [http://www.tcmagazine.com/comments.php?id=20790&catid=2.]

Pentium4 1.6A @ 1600MHz: 105s (168000 MHz-s)
Pentium4 Northwood @ 3590 MHz: 39.3s (141100 MHz-s)
Pentium4 Prescott @ 3980 MHz: 32.8s (130500 MHz-s)
Pentium4 Presler @ 4269 MHz: 31.4s (134000 MHz-s)

Pentium II Deschutes @ 504 MHz: 235s (118400 MHz-s)
Pentium III Coppermine @ 597 MHz: 194s (115800 MHz-s)
Pentium M Banias @ 1400 MHz: 76.1s (106540 MHz-s)
Pentium M Dothan @ 2949 MHz: 25.1s (74000 MHz-s)
Core Duo @ 2900 MHz: 20.6s (59700 MHz-s)
Core 2 Duo Conroe @ 4100 MHz: 12.5s (51250 MHz-s)
Core 2 Duo Penryn @ 4600 MHz: 10.4s (47800 MHz-s)
Nehalem C0 @ 5313 MHz: 7.75s (41200 MHz-s) [Fugger's recent accomplishment]

Atom @ 1600 MHz: 92.9s (148600 MHz-s)

So for SuperPi and quite a few other apps, Intel made a clock-efficiency champion out of its Pentium M line, which I remember grounded its notebook market dominance but was mostly irrelevant on the desktop because clocks were designed not to scale. Intel made some gigantic leaps in pi-calculating efficiency with a couple mobile generations. Then the breakout happened when Core 2 Duo simultaneously increased marginal efficiency and greatly increased operable frequency, leaving older architectures distantly behind on this benchmark.

AMD, on the other hand, didn't experience a huge clock-efficiency leap like the Pentium M, and many of its die shrinks were also uneventful at the core and thus insignificant impacts on SuperPi efficiency (e.g., Sledgehammer to San Diego to Toledo to Brisbane).

Though certainly not the only factor, SuperPi generally favors decent amounts of fast cache; particularly where cache is slow and/or reduced, I notice SuperPi times go up.
 

First of all, a 2.4 Venice was listed as....? A 4800, which meant? A P4 running at 4.8Ghz, not 3.2. Now, if you think Im wrong, think again

As for the Phenom doing better than i7, its in 1 of the reviews being done about i7, maybe Annands? And who cares? That wasnt my point, and you know it. My gpu slaughters i7 in super pi, so what? So a core duo is faster than a Phenom at 3.9Ghz? Not a C2D, but a core duo? And this makes sense? A Toledo at 2.8 was rated at what? Maybe 5Ghz P4? They just dont match up. And each gen isnt always better. Ask someone else this, or keep believing what you will, but Im again, not alone with this knowledge. Super pi favors Intel solutions, and like I said, 1 gen 3DMark favors ATI, nVidia the next, but it has nothing to do with real perf
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom