Sandy Bridge Tested in Asia

[ares1214]

http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.inpai.com.cn%2Fdoc%2Fhard%2F138458.htm

Or if the link doesnt work:

http://www.inpai.com.cn/doc/hard/138458.htm

+

http://translate.google.com/#

Looks like a preview of Sandy Bridge. We now know the non K-models get no overclocking basically, a suspicion from a while back. Fortunately though they got it up to 4.7 GHz air when they could adjust the multiplier. What is unfortunate though is that BD appears to only be 5-10% faster clock for clock, and leaning closer to the 5. However, people also have to take something into account. SB is 32nm and Nehalem/Lynnfield on 45nm. While thats nice and all for power consumption, also how we should look at "clock for clock. The i7 2600K stocks at 3.4 GHz, and will OC on air, to say a max of 4.7-4.8 GHz. i7 875K stocks at 2.93 GHz, and will hit 4.1-4.3 on air, generously. This means the i7 2600K is at about 70% of its max frequency, and the 875K is at almost exactly 70% of its max frequency as well. So considering overclocking, "clock for clock" would almost be like comparing the 875 K at 2.93 GHz, and the i7 2600K at 3.4 GHz, so stock speeds. Anandtech's review showed this 15% clock speed increase yield just that, 15% more performance. That added on to the 5% IPC gains, and you have this 2600K beat a i7 930/920/950/860/875K by around 20% "clock for clock". Id also like to see more benchmarks on the IGP

 

[jj463rd]

Hey thanks and kudos on the links.The 2600k looks like my next CPU for an upcoming Microsoft Flight rig.

 

[fazers_on_stun]

Hmm, the CPU-Z screenie shows a SB i5-2300 engineering sample, yet the details seem to show a 2600 model (8 MB L3, 3.4GHz, 8 threads), which is how it is listed in the benchmarks.

Anyway, I'll wait for the shipping CPUs to get benched in English so that I don't get a headache from Google translations .

 

[jobz000]

There's actually an English Version link at the top right hand corner. I only found out after reading google translated version!

 

[halfcalf]

Yeah, I'm seriously doubting that article over the confusion over the 2300 vs. 2600 as well. If those are the benchmarks for the lower CPU, then it's fine. But if that's the performance of the 2600 then it is a huge disappointment. And I don't think that Intel would issue such a ballyhooed CPU that would have such a slight performance edge over what is now available.

 

[Haserath]

What is unfortunate though is that BD appears to only be 5-10% faster clock for clock, and leaning closer to the 5.

Was that meant to be SB, not BD?

IF this is true, than BD might actually have a chance to catch up to SB this round, but I am somewhat suspicious about this being correct.

The gaming advantage isn't much, but many are graphics bound titles, except SCII, which is processor bound, showed huge improvement.

 

[eyefinity]

http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.inpai.com.cn%2Fdoc%2Fhard%2F138458.htm

Or if the link doesnt work:

http://www.inpai.com.cn/doc/hard/138458.htm

+

http://translate.google.com/#

Looks like a preview of Sandy Bridge. We now know the non K-models get no overclocking basically, a suspicion from a while back. Fortunately though they got it up to 4.7 GHz air when they could adjust the multiplier. What is unfortunate though is that BD appears to only be 5-10% faster clock for clock, and leaning closer to the 5. However, people also have to take something into account. SB is 32nm and Nehalem/Lynnfield on 45nm. While thats nice and all for power consumption, also how we should look at "clock for clock. The i7 2600K stocks at 3.4 GHz, and will OC on air, to say a max of 4.7-4.8 GHz. i7 875K stocks at 2.93 GHz, and will hit 4.1-4.3 on air, generously. This means the i7 2600K is at about 70% of its max frequency, and the 875K is at almost exactly 70% of its max frequency as well. So considering overclocking, "clock for clock" would almost be like comparing the 875 K at 2.93 GHz, and the i7 2600K at 3.4 GHz, so stock speeds. Anandtech's review showed this 15% clock speed increase yield just that, 15% more performance. That added on to the 5% IPC gains, and you have this 2600K beat a i7 930/920/950/860/875K by around 20% "clock for clock". Id also like to see more benchmarks on the IGP

That's ridiculous even by your standards ares.

From that link, Sandy Bridge is 6% faster clock for clock (I didn't check the gaming benchmarks). That means it's going to be 6% faster at the same clock speed and no dressing it up different. I thought all nehalem i7's overclocked to 4.5ghz with ease anyway? I guess now that Sandy Bridge isn't overclocking all that higher, it's safe to lower that to 4.2ghz "generously".

I don't see a lot of people queuing up to replace their desktop nehalems with this. It should be better in a notebook as clearly lower tdp is where the extra has gone, not into extra performance. This shouldn't surprise anyone, notebooks are the biggest market.

 

[sportsfanboy]

^ I'll admit I'm lazy and didn't read the article. Where did all those 30% faster claims come from? Or Sandy Bridge is the biggest performance increase since core2 was released. I'm not a fanboy(hense the name), but getting AMD back into the performance game seems like a good thing, from a buyers standpoint.

 

[fazers_on_stun]

^ I'll admit I'm lazy and didn't read the article. Where did all those 30% faster claims come from? Or Sandy Bridge is the biggest performance increase since core2 was released. I'm not a fanboy(hense the name), but getting AMD back into the performance game seems like a good thing, from a buyers standpoint.

I wouldn't put too much stock in the article, considering the incorrect CPU-Z screenshot. There are other previews out (i.e., 2 articles on Anandtech) that show an i5-2400 with no working turbo keeping up with an i7-980X in some games, and on average about 20% faster clock-for-clock over Nehalem. Even JF-AMD who works for AMD mentioned in another thread here that he's seen reviews going from 15-30% higher performance over Nehalem.

But we will all know the truth sometime around January 9th it appears. In fact we probably won't see anything else besides SB reviews that day . And we should see the desktop version of Bulldozer in April I think..

 

[ares1214]

That's ridiculous even by your standards ares.

From that link, Sandy Bridge is 6% faster clock for clock (I didn't check the gaming benchmarks). That means it's going to be 6% faster at the same clock speed and no dressing it up different. I thought all nehalem i7's overclocked to 4.5ghz with ease anyway? I guess now that Sandy Bridge isn't overclocking all that higher, it's safe to lower that to 4.2ghz "generously".

I don't see a lot of people queuing up to replace their desktop nehalems with this. It should be better in a notebook as clearly lower tdp is where the extra has gone, not into extra performance. This shouldn't surprise anyone, notebooks are the biggest market.

You and Dipankar never fail to impress. You have a valid point, clock for clock, it isnt much better, but it wasnt really supposed to be. For all intents and purposes, SB is just a die shrink, added better IGP, and some tweaks here and there. Like if HD6000 was on 32nm. Ivy Bridge is the one that will bring the rain, and if that isnt much faster clock for clock, then we can call it a disappointment. However, the general reason we compare things clock for clock is to show given the same overclocking headroom, how do they perform. You really think people care about "clock for clock" its just a way to show a "fair" comparison. If bulldozer is clocked at 75 GHz, and Ivy has 16 cores, will it be fair to compare clock to clock? Core to core? Not really. The point is, IF SB OC's almost the exact same amount from stock as i7, i personally considering the stock of SB "clock for clock" with the stock of the i7's. Whats the point of comparing clock to clock if the stock is 20% higher as is the overclocking headroom? Think what you will, but you lose this one.

Dipankar, until i see prices, im not overly impressed with SB. If Intel prices the K series much higher than the normal ones, its a failiure for me. Id hope the 2500 would be atleast down to $200, with the 2600K no higher than $329. If its $500+ like intel loves to do, i for one will be most unimpressed. It is most definitely too early to say this will smash Bulldozer. For one, this likely, or hopefully, wont really compete against BD, IB will. Secondly, BD is the first time AMD has really put their thinking caps on since K7, remember what happened that time?

 

[Haserath]

Dipankar, until i see prices, im not overly impressed with SB. If Intel prices the K series much higher than the normal ones, its a failiure for me. Id hope the 2500 would be atleast down to $200, with the 2600K no higher than $329. If its $500+ like intel loves to do, i for one will be most unimpressed. It is most definitely too early to say this will smash Bulldozer. For one, this likely, or hopefully, wont really compete against BD, IB will. Secondly, BD is the first time AMD has really put their thinking caps on since K7, remember what happened that time?

Second thread on this is also up with rumored prices:
http://www.tomshardware.com/forum/294352-28-latest-sandy-bridge-news#t2206209

 

[ares1214]

Second thread on this is also up with rumored prices:
http://www.tomshardware.com/forum/294352-28-latest-sandy-bridge-news#t2206209

Why thank you. If those prices are real, then this is fairly decent, not spectacular, but i dont think it was meant to be. A bit disappointing the 2500 dont have HT and you need to spend $300 for HT, i was hoping the 2500 would, that would make it a steal. But i think it might be the 2500K for me, unless BD comes sooner.

 

[BadTrip]

You and Dipankar never fail to impress. You have a valid point, clock for clock, it isnt much better, but it wasnt really supposed to be. For all intents and purposes, SB is just a die shrink, added better IGP, and some tweaks here and there. Like if HD6000 was on 32nm. Ivy Bridge is the one that will bring the rain, and if that isnt much faster clock for clock, then we can call it a disappointment. However, the general reason we compare things clock for clock is to show given the same overclocking headroom, how do they perform. You really think people care about "clock for clock" its just a way to show a "fair" comparison. If bulldozer is clocked at 75 GHz, and Ivy has 16 cores, will it be fair to compare clock to clock? Core to core? Not really. The point is, IF SB OC's almost the exact same amount from stock as i7, i personally considering the stock of SB "clock for clock" with the stock of the i7's. Whats the point of comparing clock to clock if the stock is 20% higher as is the overclocking headroom? Think what you will, but you lose this one.

Dipankar, until i see prices, im not overly impressed with SB. If Intel prices the K series much higher than the normal ones, its a failiure for me. Id hope the 2500 would be atleast down to $200, with the 2600K no higher than $329. If its $500+ like intel loves to do, i for one will be most unimpressed. It is most definitely too early to say this will smash Bulldozer. For one, this likely, or hopefully, wont really compete against BD, IB will. Secondly, BD is the first time AMD has really put their thinking caps on since K7, remember what happened that time?

Don't let jenny and dips**t discourage you. Good things are coming from both AMD and Intel. Good Times for All!!!

 

[halfcalf]

Good things are coming from both AMD and Intel. Good Times for All!!!

2011 might be a banner year for CPUs unlike any we've seen lately. Let's hope and pray!