[NEW STUFF] Intel LGA1155

[superrugal]

Sandy Bridge vs Core i7 in Cinebench 10... 次世代PC向けCPU『Sandy Bridge』でベンチマーク！

Does Cinebench 10 use AVX? Or is that speedup due to IPC improvements?

I think the frequency is different.

[Hornet331]

hard to make any assumptions, since we don't know the sepcs.

[mAJORD]

The SB renders it about twice as fast

it's scoring about 20000. The i7, about 10000, so all we can assume is the i7 is under 2Ghz? :|

[Kimi]

Sandy Bridge vs Core i7 in Cinebench 10... 次世代PC向けCPU『Sandy Bridge』でベンチマーク！

Not knowing the test platforms and CPU frequencies that test means nothing.

[massman]

yeah i know but nobody actually uses it... at least nobody did when i last checked

here, there isnt a single ddr3 800 listed on geizhals:
http://geizhals.at/deutschland/?cat=ramddr3regecc

and i cant find a single ddr3 800 dimm on newegg either...
so you actually have to DOWNCLOCK your memory BELOW the slowest available speed to use 3 dimms per channel... ouch...
they could as well have added that they support 4 dimms with ddr3 533... its like... oh... really?... thats... err... cool!... i guess...

http://geizhals.at/deutschland/a533042.html ?

[OhNoes!]

@ 0:16 Seconds....[Presenter]:"The other one... is top of the line i7 right? (voice in background - "yes"). Which is the best quadcore, hyperthreaded.... let's say the best comparison we have today compared to Sandybridge....."

i7 975? If so, Sandybridge is going to be a beast. This should also put all the ipc arguments to rest, unless this particular chip is clocked well above 4GHZ and Intel managed to increase HT efficiency over 50%.

[G.Foyle]

1) Cinebench R10 used
2) no word about clock speeds - could be 2 GHz i7 vs 3 GHz SB for all I know
3) I was under the impression that "IPC" doesn't imply a single thread perfromance. Maybe HT enhancements = IPC boost?

[mAJORD]

@ 0:16 Seconds....[Presenter]:"The other one... is top of the line i7 right? (voice in background - "yes"). Which is the best quadcore, hyperthreaded.... let's say the best comparison we have today compared to Sandybridge....."

i7 975? If so, Sandybridge is going to be a beast. This should also put all the ipc arguments to rest, unless this particular chip is clocked well above 4GHZ and Intel managed to increase HT efficiency over 50%.

A 975 doesn't take 1:30 to get through CB10, not evne close, so we're still at square one

[OhNoes!]

A 975 doesn't take 1:30 to get through CB10, not evne close, so we're still at square one

Good point, I'll do a quick test.

Edit: i7 @ 3.33GHZ gave me 55 seconds in CB 10 with a score of 16,042

The i7 test took about 120 seconds in the video. i7 @ 2.40GHZ gave me 115 seconds. Hmmm, interesting.

[saaya]

http://geizhals.at/deutschland/a533042.html ?

you want to put this in a server? are you sure?

i knew youd link to one of those 2 ddr800 non ECC kits listed on geizhals :P heheh
what does that prove?
ddr3 800 is pretty much a dead standard, ffs, intel even told everybody so themselves behind closed doors last year... back when they didnt officially support ddr3 1333 for 1366, they said for 1156 they are canceling ddr3 800 qualification resources since nobody is using it and its a dead standard, and they would move resources to ddr3 1333 and 1600 qualification on 1156...

the only segment you will actually find ddr3 800 memory is in laptops... they only added the support info for ddr3 800 to distract from the ridiculous 1 stick per channel support for speeds people actually use... i cant believe people actually defend them for a lame marketing move like this...

Good point, I'll do a quick test.

Edit: i7 @ 3.33GHZ gave me 55 seconds in CB 10 with a score of 16,042

The i7 test took about 120 seconds in the video. i7 @ 2.40GHZ gave me 115 seconds. Hmmm, interesting.

lol, 2.4ghz i7... "the fastest i7 chip" ... hmmm wait a minute... the first SB parts will launch for laptops, wont they? then this actually makes sense!

Core i7 = 73s
SB = 44s
thats a 65% boost... if the i7 was at 2.4ghz then the SB was running at 3.9ghz... heh, not possible...

the fastest mobile i7 quadcore is a 920XM running at 2ghz... that would put SB at around 3.3ghz... sounds a lot more reasonable... the 920xm has a 55w tdp, so 3300mhz for a SB chip in 55W sounds perfectly fine...

as to how a 2ghz i7 crunches through cinebench in 73s, well its probably with turbo on, and on a well cooled platform so its speeding up to... idk, how high does the 920xm turbo? intel removed the turbo multi information from their cpu spec site

i guess it goes up to 2300mhz?
that means the sb chip probably had turbo on as well...

if this is true, then intel deserves an epic facepalm award...
once again they pulled out a dusty ancient product to compare it to their new part, as its just so damn hard to make their new product look any good otherwise... lame! :P the 920xm is a 45nm portable waterboiler and hair dryer...

[xVeinx]

3) I was under the impression that "IPC" doesn't imply a single thread perfromance. Maybe HT enhancements = IPC boost?

Yes, SMT is being considered here in terms of IPC boost (and is the area where SB has the most room to grow).

[saaya]

that would make sense... so of the 65% boost 10-20% could be from more efficient SMT...
which means the frequency difference would only be around 50%, 2ghz vs 3ghz or 2.4ghz vs 3.6ghz.
and ~3.6ghz is the rumored max turbo speed for SB chips, so that would make perfect sense then...

[informal]

Anyone else remembers the Yorkfield test with "hacked" Cinebench engine (they used newer Cinema engine inside the older Cinebench benchmark software)? Yorkfield scored a lot higher this way when compared to Kenstfield doing the runs in Cinebench ,when in reality it was just a hair faster in *real* Cinebench than Kenstfield at the same clock . This presentation was done ,IIRC, at one of the IDFs and it was only later on when it was discovered they used newer engine to benchmark the 45nm shrink of the Core.

Also sayya has a good point,there is a chance they just used mobile versions of the chips.SB is 32nm,meaning higher default and turbo clocks than 45nm Nehalem QC with SMT.

EDIT:
actually it was a Nehalem presentation which had the "hacked" Cinema engine used,not Yorkfield.My mistake. It's in this thread.

[superrugal]

I guess SandyBridge has an improved Turbo Mode

[terrace215]

socket 2011: new multiprocessor(2-4 cpu on one motherboard without a cpu extension board) socket to replace socket 1567(which was annonced 3 month ago so we wont see socket 2011 until 1 or 2 year i think)

socket 1356: new uni processor and dual processor socket to replace socket 1366 ( this is sandy bridge b2 that everyone here is waiting for and should come around q3 2011)

socket 1155: new uni processor socket to replace low end socket 1156 ( first sandy bridge cpu that we will see will be on this socket)

S-2011 is due Q2/Q3 of next year. Unclear if S-1356 has been cancelled, with those parts going S-2011 instead. (One can hope)

[demonkevy666]

I can read the cinebench on 720P on youtube.
Sandybridge scores 19,843 looks like it's at 2.99ghz on the screen.
I7 2.93ghz. 14,961

[xdan]

Sandy Bridge might be 25-30% more performant clock per clock than Nehalem in some syntetic benchmarks, what is so hard to believe. And that probably with only 6mb L3 cache instead of 8MB used on Nehalem.
I mean that was the overall the difference between Yorkfield and Nehalem.
There are two years now from Nehalem, so i think that it's not so strange for Intel to boost performance with 20-30% again.
I mean Lynnfield, Clarkdale, where all derivates from Nehalem.
SB will be something more advanced, more higher IPC, ....
I mean Buldozer is coming........ . Intel must be prepared.
Anyway if SB will be 25-30% more performant than Nehalem, than Buldozer might be doomed. I don't think that Amd is capable to make a jump of 50% to make Buldozer 50% more performant than Phenom II and say to equal SB.

[informal]

Sandy Bridge might be 25-30% more performant clock per clock than Nehalem in some syntetic benchmarks, what is so hard to believe. And that probably with only 6mb L3 cache instead of 8MB used on Nehalem.
I mean that was the overall the difference between Yorkfield and Nehalem.
There are two years now from Nehalem, so i think that it's not so strange for Intel to boost performance with 20-30% again.
I mean Lynnfield, Clarkdale, where all derivates from Nehalem.
SB will be something more advanced, more higher IPC, ....
I mean Buldozer is coming........ . Intel must be prepared.
Anyway if SB will be 25-30% more performant than Nehalem, than Buldozer might be doomed. I don't think that Amd is capable to make a jump of 50% to make Buldozer 50% more performant than Phenom II and say to equal SB.

We have SB benchmarked already... Except in a few niche tests(synthetic BW tests inside everest),there is hardly a difference between SB and i7. The last video is full of unknowns so it's not a best "proof" of 25% IPC claim. From what we have seen in the real SB review @ coolaler,software is still not ready for SB's main weapon and that's AVX.

[xdan]

Sandy Bridge might be 25-30% more performant clock per clock than Nehalem in some syntetic benchmarks

Will see anyway you could judge this from another point to, is SB has 6mb L3 cache this means 25% less cache than Nehalem, so same performance or better with 25% less L3 cache this means something.
What i expect from SB is much lower prices i mean 229$ for 4core HT and 169$ for 4cores no HT, i mean really mainstream.
In that case i don't know what Amd can do.

[Manicdan]

We have SB benchmarked already... Except in a few niche tests(synthetic BW tests inside everest),there is hardly a difference between SB and i7. The last video is full of unknowns so it's not a best "proof" of 25% IPC claim. From what we have seen in the real SB review @ coolaler,software is still not ready for SB's main weapon and that's AVX.

i think the video might have compared performance differences at the same wattage. an i7 at whatever reduced frequency to match the power consumption of the SB. otherwise i cant see any real valid case that was tested.

[FlanK3r]

I can read the cinebench on 720P on youtube.
Sandybridge scores 19,843 looks like it's at 2.99ghz on the screen.
I7 2.93ghz. 14,961

32 bit or 64bit? Bit whats coolaler ebchmarks? He benchmarked Cinebench too (R11.5, is scaling better without lags loading cores) and it will very simillary

[informal]

Will see anyway you could judge this from another point to, is SB has 6mb L3 cache this means 25% less cache than Nehalem, so same performance or better with 25% less L3 cache this means something.
What i expect from SB is much lower prices i mean 229$ for 4core HT and 169$ for 4cores no HT, i mean really mainstream.
In that case i don't know what Amd can do.

Don't forget AMD will have its own 32nm designs ready by the time SB becomes a noticeable factor in intel's product mix. I'm referring here to 2/4 core Lanno products(will crush SB in GPU part and be fast enough in CPU part,although slower) and 2/4 module Orochi which will go after mid and high end segment.As you can expect ,4 core Llano and 4 core(2 module) Orochi will be perfect to address the SB models you have mentioned,both from production cost and consumer POV.

[demonkevy666]

32 bit or 64bit? Bit whats coolaler ebchmarks? He benchmarked Cinebench too (R11.5, is scaling better without lags loading cores) and it will very simillary

64 bit.

there is something is strange about the sandy bridge load >_>.

[superrugal]

R11.5 is here

[Florinmocanu]

From the CB11.5 score, i cannot see a proper improvement over nehalem. 4.63 at 2.5 ghz vs 5.06 at 2.8ghz. Almost like the same CPU.