SB unfair in comparison reviews

[Bearach]

I can see the OP's point on some level, in that he wishes to see how much of an improvement clock for clock it is with his i5. But I also disagree, it's not an unfair comparison. You should see stock vs. stock processors.

 

[combust3r]

http://en.inpai.com.cn/doc/enshowcont.asp?id=7947&pageid=7709

i5 2300 (they've downclocked i5 2500K) is faster up to 20% (depends on the tests).

 

[Bearach]

http://en.inpai.com.cn/doc/enshowcont.asp?id=7947&pageid=7709

i5 2300 (they've downclocked i5 2500K) is faster up to 20% (depends on the tests).

Thanks for that link, seems many of the benchmarks show no noticeable difference though.

 

[Markfw]

but then you should bench the Sandy Bridge cpu at a speed of 4400MHz at least. The OP wants the SB benched stock vs. an overclocked cpu. I can see the point of wanting that info, just for forum talk 'n all, but how does it help in real life? If one is OK overclocking older cpus, they are likely OK with overclocking the new cpu.

Isn't that what I said ?? I did not add actual speeds, since I don;t have the chip, and I don't know for sure what they will cost.

 

[cusideabelincoln]

Now of course there *is* a purpose for a clock for clock comparison between SB and old gen i5/i7s to see how the uArch compares, but thats purely for academic purposes and doesn't portray how these CPUs perform in the real world, either 'out of the box' or at their highest potential overclocks.

Basically this. It's nice to know so that we can put on our thinking caps, but you also can't ignore Sandy Bridge does indeed have another advantage and that advantage is clockspeed.

There obviously is an improvement in the architecture. If you are to compare the 2500k, with no hyperthreading and 6MB of L3, to Lynfield, with no hyperthreading and 8MB of L3, the 2500K still ends up on top clock for clock. And most reviews show the i5 2500k, with no hyperthreading, going toe to toe and outpacing past Core i7s, despite them having hyperthreading.

Like here, the 2500K with 2MB less L3 and no hyperthreading and less memory bandwidth is generally outpacing the i7 975 Extreme:
http://www.anandtech.com/bench/Product/288?vs=99

So obviously SB has a more efficient architecture. And I bet we'll really see these things shine on socket 2011, with higher memory bandwidth and more cache.

 

[Hogan773]

How is that 'unfair'? You either compare products at their shipping clockspeeds (which is what most reviewers have done), or you do an overclocking comparison with both CPUs overclocked to the max. You don't overclock one and leave another at stock... now THAT would be unfair.

Now of course there *is* a purpose for a clock for clock comparison between SB and old gen i5/i7s to see how the uArch compares, but thats purely for academic purposes and doesn't portray how these CPUs perform in the real world, either 'out of the box' or at their highest potential overclocks.

+1. OP is off base. His logic is that just because he can OC his i5 750, the performance difference between his i5 and STOCK SB isn't as large as shown in the reviews. So what? If I buy a SB I can OC, very likely to a higher level than the i5 750, and then the performance difference will be as big or even bigger. If SB was locked down from overclocking then I'd agree with his logic, but it ain't!

 

[dualsmp]

1680x1050 also tends to be the standard "CPU review" resolution since it shows more dramatic results between CPU's in games. Typically no AA is used either. Throw in 1920x1200 with AA cranked up and the results don't look nearly as dramatic between CPU's.

 

[Tsavo]

Actually, I think there are two frames of mind her, and both are valid.

IF you are an overclocker, you want to see what both chips can REASONABLY do on a consistant basis, then bench them at those speeds, comparing chips of like price (including platform).

If you DON'T overclock, then take chips of like price(including platform) and compare them.

At least those are the 2 views that I see, but both are perfectly valid.

This.

Clock for clock comparisons are valid.
Overclock comparisons are valid.
Price point comparisons are valid.

Beyond all that, the SB wins because not only because of the higher stock clocks, but a higher OC envelope at a much lower power draw.

 

[IntelUser2000]

All of the reviews I've seen so far that include an i5- 750 seem to forget that the 2500K is over 600mhz faster at stock speed. I'd like someone to OC the 750 to 3.3Ghz and see how much advantage the SB really has. Has anyone seen a review like this yet?

http://www.computerbase.de/artikel/...y-bridge/47/#abschnitt_turboskalierungsrating

Just hover your mouse over the 760 and it will show the differences.

Solid 15% per clock with the 2500 and 17% with 2600.

 

[llee]

bear in mind that you can't overclock as high on SB as you could with Lynnfield

 

[Nemesis 1]

All of the reviews I've seen so far that include an i5- 750 seem to forget that the 2500K is over 600mhz faster at stock speed. I'd like someone to OC the 750 to 3.3Ghz and see how much advantage the SB really has. Has anyone seen a review like this yet?

Do all the comparsions ya want . The reviews sites have been using dollar for dollar comparsions for along time now to help AMD out . Now ya want them to suddenenly change and say Its clock for clock . They would look pretty bad ,

As it is They comparred to dollar for dollar all the way up to the big daddy . how dare they . The only right way to compare is base clock to base clock . Any performance gain is the increased performance achieved. Its nice you want the reviewers to do things in a manner that would fit in your world but that world is not real its imagined the facts speak for themselves. Base clock to base clock SB kicks but. Than overclocking is great . Some will get bad cores that do only 4.6 GHZ i hear. But I would like a shot at those cpus.

 

[Nemesis 1]

bear in mind that you can't overclock as high on SB as you could with Lynnfield

Got a link for this. Surely you know of the ceral special K

 

[perdomot]

Quite a lot of feedback so I'll try to reply as much as I can. First, thanks to the folks for the links to help give me an idea of what the new cpus really bring to the table. Second, perhaps "unfair" isn't quite the best way to say it but when you have all these sites going "WOW, look how much better the new cpus are!" and you have such a large difference in megahertz, it gets rather irksome. A lot of people like me recently switched over to Nehalem and want to know if it was a big mistake or not and its kind of hard when the compared cpus are 600+ megahertz apart. As for how overclocking figures in, its all a gamble although plenty of the 750/760 have been showing getting 4.2-4.4Ghz which is the same range SB is in. All in all, most reviews have commented that SB is basically bringing performance that the more expensive cpus had down to mainstream pricing which is always a good thing. Thanks everyone.

 

[RussianSensation]

i5 750 @ 3.33ghz vs. i5-2500k
http://www.semiaccurate.com/2011/01/03/intel-core-i7-2600k-review/

 

[perdomot]

i5 750 @ 3.33ghz vs. i5-2500k
http://www.semiaccurate.com/2011/01/03/intel-core-i7-2600k-review/

Thanks for the link. The conclusion of the article pretty much reflects what I suspected. That is that if you have a good Core i5 rig that is nicely OC'd, SB offers some advantage in certain scenarios but probably not enough to justify upgrading.

 

[Makaveli]

I thought the Hardocp reviews good also

http://www.hardocp.com/article/2011/01/03/intel_sandy_bridge_2600k_2500k_processors_review/1

showing x980 @ 4.4 vs 2600k @ 4.4

also an 920 @ 3.6 was pretty much equal to the 2600k stock.

 

[Rifter]

I thought it was common knowlege that if you have a high clocked i5/i7 that SB really isnt much of an upgrade.

 

[Makaveli]

common sense ain't so common these days

 

[Tsavo]

bear in mind that you can't overclock as high on SB as you could with Lynnfield

Define "clock as high". Clock as a % increase over their base or the max GHz they can hit? Show me a Lynnfield that can clock to 4.2 GHz as breathlessly as an SB can and you may have a case.

At this point in time, there are no SB's in consumer hands, so who know what a safe OC will be before cranking the vcore. 4.2 GHz? 4.4 GHz?

One stands a much better chance at hitting 4.2 GHz with a SB than with a Lynnfield, at a drastically lower power consumption to boot.

The Lynnfields were great chips when released and they still are, but they've been outmoded by a product with the same price that has a much higher performance ceiling.

I've got a 750 @ 3.6 GHz and that's as far as I can go before I start twiddling with the Vcore. SB easily adds another 600 MHz to that, on top of the 10-15% boost in performance the SB has all by itself.

I'm not going to run out and buy a SB to replace my poor old Lynnfield, but if the s2011 parts are any bit as good as the 1155 parts are, I'll be all over it with a nice G3 or C400 to go along with it.

 

[RussianSensation]

I've got a 750 @ 3.6 GHz and that's as far as I can go before I start twiddling with the Vcore. SB easily adds another 600 MHz to that, on top of the 10-15% boost in performance the SB has all by itself.

I bet you can do 3.9-4.0ghz at 1.25 VTT and 1.312V/1.325V. Both are 100% safe on Lynnfield. If my 860 can do with HT, there is no reason your 750 can't. Time to revisit that overclock Tsavo

 

[Nemesis 1]

So you argue that his will do better . But you take it as gosphel that SB 2500k/2600K have trouble reaching 4.7GHz . Were is this great pool of information and comparisons that lead ya to this conclusion.

Lets not forget about untested as yet apps . That need to be recompiled.

http://www.guru3d.com/article/core-i5-2500k-and-core-i7-2600k-review/17

 

[Tsavo]

I bet you can do 3.9-4.0ghz at 1.25 VTT and 1.312V/1.325V. Both are 100% safe on Lynnfield. If my 860 can do with HT, there is no reason your 750 can't. Time to revisit that overclock Tsavo

First of all, my mainboard has a Foxconn socket, so pushing for 4 GHz + is just asking the Fire Department over for a BBQ.

I do not doubt at all your suggestion that small increases in v can stick me at 3.8 GHz or better, but after 3.8 the Lynnfield really starts consuming power and making noise. I'm not fond of either, so 3.6 GHz is my last stand on this CPU. For others it will be different, but for me, I prefer a modicum of silence to go with my vast cosmic dark.

That's why I am now finding s2011 so appealing. I can add another 1 GHz ++ to my box without adding any more heat or noise which to me is a huge benefit. Add to that SATA 3 supported natively on an Intel chipset and the next gen SSD's doing 415 MB reads and 260 MB writes...and yes, I 'm there.

 

[Fallen Kell]

bear in mind that you can't overclock as high on SB as you could with Lynnfield

So 4.6-4.8GHz on AIR isn't as high as Lynnfield? What are you smoking?

 

[cusideabelincoln]

I thought it was common knowlege that if you have a high clocked i5/i7 that SB really isnt much of an upgrade.

Just like it's been for the last few iterations...

If you had a high-clocked Kentsfield then Penryn (at stock) wasn't much of an upgrade.
If you had a high-clocked Penryn then Nehalem/Lynnfield (at stock) wasn't much of an upgrade.

And like those previous generations, there were always some exceptions where the new chip did some things much better that the clockspeed of the previous gen couldn't compensate for. But if you overclock the new gen, then it starts the shine (when compared to an overclock of the previous gen). Same thing with Sandy Bridge.

 

[IanWorthington]

Just wondering about my old 4040 at 3.3GHz.