Early SandyBridge benchmarks (translated)

S

seanmcd

[H]ard|Gawd
Joined
Jan 20, 2003
Messages
1,457
This is from the site "inpai.com.cn" (use the better English links below)
i7 2600k
http://translate.google.com/transla.../www.inpai.com.cn/doc/hard/138458.htm&act=url

i5 2500K
http://translate.google.com/transla...u=http://www.inpai.com.cn/doc/hard/139753.htm

Sorry if this has been posted. But a very interesting look at the new 1155 cpu.

EDIT: Here are direct links to the English version of the site (thanks harpoon!)

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

http://en.inpai.com.cn/doc/enshowcont.asp?id=7944
 
Last edited:
My takeaway: At 1680x1050, games seem to be GPU limited and don't benefit much from the improvements of SB. Other applications tend to, however.

I couldnt find any overclocking benchmarks in that article. They do show it completing SuperPi at 4.7ghz
 
Based on the articles listed in reference, it appears

1. At least you get visible power efficiency improvement for those models transition from 45nm to 32nm.

2. However, if you intend to buy cheapest model to overclock to extreme rate, looks like this does not work out too well, for now, as appears that route is severely restricted.

3. So compare this for a fictional example, say you buy SB 2.8GHz 2C/2T 32nm, due to architectural improvement, it will definitely be faster than 45nm C2D 2.8GHz 2C. However, since this low-end inexpensive model does not allow easy overclock, it does have alternative moments when facing the 45nm C2D overclock to 4.0 GHz(examples, could be any higher clocks above 3.4Ghz)

4. But under any reasonable circumstances, the 32nm SB will have power efficiency.

So its your choice, want supreme overclock, pay for the K-variant SB, want cheapest overclocking performance but don't care about power, current 1156 core i-series. Already a 45nm C2D owner --> thinking time :)

5. You do get newer incarnation of Intel GPU in SB, if integrated GPU is important.

Above is all based on the posted article references. As SB is still not announced, this is unsubstantiated opinion, not facts.

Edit : another thing is to wait for motherboard makers to see if they provide any surprise in overlocking potential on low-end non-K SB models.
 
There were rumors of stable overclocks on air @ 5Ghz, if true what do you think the upper end will be with extreme cooling (e.g. water)?
(referring to K series)
 
loki7 said:
There were rumors of stable overclocks on air @ 5Ghz, if true what do you think the upper end will be with extreme cooling (e.g. water)?
(referring to K series)
Click to expand...


all depends but i'd expect something in the mid to high 6ghz range on LN2. hasnt been to many processors that can break the 7Ghz barrier on LN2 that i've seen. liquid helium on the other hand they have been able able to break 7Ghz but not 8Ghz. so it should be interesting. if they can hit 8Ghz on liquid helium that would be insane. but over all the thing to worry about is how well it scales with voltage. ya may be able to get 6ghz but it might require 1.6v to do it. so who knows at this point.
 
loki7 said:
There were rumors of stable overclocks on air @ 5Ghz, if true what do you think the upper end will be with extreme cooling (e.g. water)?
(referring to K series)
Click to expand...

From their 2600k article, they got 4.7ghz with 1.38 but it doesn't seem they verified stability. Still, if early ES silicon can hit that then it would seem 5ghz air on a good chip should be attainable.
 
They are testing with unlocked ES chips manually clocked to the retail speeds for comparison, hence my comment about early silicon. Retail 2600K chips should overclock significantly better than these early samples.
 
R-Type said:
They are testing with unlocked ES chips manually clocked to the retail speeds for comparison, hence my comment about early silicon. Retail 2600K chips should overclock significantly better than these early samples.
Click to expand...

Interesting posts about these "unlocked" multipliers. Each chips seem to hit a wall at specific multiplier that no amount of voltage or minor tweaking can surpass. Retail chips seem to exhibit this as well as Eng Samples.

http://www.xtremesystems.org/forums/showpost.php?p=4665104&postcount=40

http://www.xtremesystems.org/forums/showpost.php?p=4665258&postcount=54
 
I smell a lockdown for overclockers...........better get the $£$£$£$£$ out :(
 
Ehren8879 said:
Interesting posts about these "unlocked" multipliers. Each chips seem to hit a wall at specific multiplier that no amount of voltage or minor tweaking can surpass. Retail chips seem to exhibit this as well as Eng Samples.

http://www.xtremesystems.org/forums/showpost.php?p=4665104&postcount=40

http://www.xtremesystems.org/forums/showpost.php?p=4665258&postcount=54
Click to expand...

Odd, reminds me of the inifinite wall at 4ghz for the early phenom II's in 64bit mode.

It will be interesting to see how this holds for a wider install base of retail samples.

I have seen testing with the bus speed up to 106mhz, which could combine with a 47 or 48 multiplier to get the 5ghz mark.

Should be interesting :D
 
R-Type said:
Odd, reminds me of the inifinite wall at 4ghz for the early phenom II's in 64bit mode.

It will be interesting to see how this holds for a wider install base of retail samples.

I have seen testing with the bus speed up to 106mhz, which could combine with a 47 or 48 multiplier to get the 5ghz mark.

Should be interesting :D
Click to expand...

Someone posted a shot a few weeks ago of a 2600K (I believe it was) with a 50 multiplier. I can't believe we are back to the days of overclocking the PCI bus.
 
Forceman said:
Someone posted a shot a few weeks ago of a 2600K (I believe it was) with a 50 multiplier. I can't believe we are back to the days of overclocking the PCI bus.
Click to expand...

Trouble with Sandy Bridge is other clocks are tied to the base clock. Overclocking the base clock, even slightly now drives these other clocks out of stable range. Hence, SB's need for multiplier overclocking for all processors.

Too bad there isn't a way to put these other clocks on a divider... or will there be?
 
Ehren8879 said:
Trouble with Sandy Bridge is other clocks are tied to the base clock. Overclocking the base clock, even slightly now drives these other clocks out of stable range. Hence, SB's need for multiplier overclocking for all processors.

Too bad there isn't a way to put these other clocks on a divider... or will there be?
Click to expand...

If there is a way, the motherboard manufacturers will find it.
 
Dan_D said:
If there is a way, the motherboard manufacturers will find it.
Click to expand...
yup.

Ehren8879 said:
Interesting posts about these "unlocked" multipliers. Each chips seem to hit a wall at specific multiplier that no amount of voltage or minor tweaking can surpass. Retail chips seem to exhibit this as well as Eng Samples.
Click to expand...
Remember that each multi is 100Mhz. Most current chips will hit a wall at some point too if you use 100Mhz increments.
 
Hmmm, too early to tell. could be a BIOS issue or something. Either way i'd be happy with 4Ghz+ on these chips :)
 
4.5-5ghz likely be the norm with overlocking these.
I aim to get a asus sabertooth and a 2500k.
Budget overclocking is here ;)
 
Flopper said:
4.5-5ghz likely be the norm with overlocking these.
I aim to get a asus sabertooth and a 2500k.
Budget overclocking is here ;)
Click to expand...


I think 2600K is an entirelty new platform. So will it work on your Asus Sabertooth X58 which is an old LGA 1366 platform?
 
Yeah there's a P67 version, it's the one with everything except the slots covered in a plastic shell.
 
You must log in or register to reply here.
Back
Top