The Sandy Bridge Preview (Anand)

[terrace215]

On the Llano side most likely yes, on the BullDozer side its pointless to speculate at this point. BD is completely different from Phenom II as you know. Its anyones guess what and when it will bring to the show.

What I'm saying is: Consider the current single-threaded IPC differential between Ph-II and Nehalem, then look at what SB just brought over Nehalem, and what BD is likely to bring over Ph-II in single-thread IPC (its weakest point)...

Looks like the gap remains about where it is. (SB over BD ~= Nehalem over Ph-II)

And for client, then consider that by the time BD launches for it, Intel will likely be shipping the *mainstream* versions of Ivy Bridge at 22nm!

[Hornet331]

I like how everyone just sees the looking as a money grabbing scheme, while totaly neglecting the technical pov.

It's a design decion they made, to satisfy the oem to save money on the boards. I am sure they could have made another desgin which allowes you clocking, even with the ondie clock generator, but i am also sure such a desgin would need a lot more validation and also increases the design complexity or the robustness.

Its easier to desgin a part that operates at a specific conditions then over a wide area.

[informal]

What I'm saying is: Consider the current single-threaded IPC differential between Ph-II and Nehalem, then look at what SB just brought over Nehalem, and what BD is likely to bring over Ph-II in single-thread IPC (its weakest point)...

Looks like the gap remains about where it is. (SB over BD ~= Nehalem over Ph-II)

And for client, then consider that by the time BD launches for it, Intel will likely be shipping the *mainstream* versions of Ivy Bridge at 22nm!

Yeah they are DOOMED!1!1!
How about we wait and see them compared,side by side,out of the box in client workloads before making doomsday scenario prediction

[Andypro1]

Hi,

I would just like to chime in here and speak to some of the comparisons I have been seeing with regards to Sandy Bridge's IGP performance and Bulldozer's 480sp IGP. The Sandy Bridge sample shown by Anandtech is the low-end version with only 1 "core" or 6 Intel-type-SP-unit-things. The problem I have trying to speculatively compare that against Bulldozer's 480sp wonderkind is that you're comparing a low-end mainstream chip that will be out in late Q4 2010 to a high-end chip that won't be out until sometime in 2012 (in my opinion - I know there's no hard date for Bulldozer). There is enough trouble trying to compare SPs from differing architectures - we need to keep the market segment and timeframe at least somewhat in line if we're to make any reasonable comparison. If you were to guess how many SPs will be on a high-end Ivy Bridge when Bulldozer arrives, I would presume you wouldn't guess "6."

[informal]

Hi,

I would just like to chime in here and speak to some of the comparisons I have been seeing with regards to Sandy Bridge's IGP performance and Bulldozer's 480sp IGP. The Sandy Bridge sample shown by Anandtech is the low-end version with only 1 "core" or 6 Intel-type-SP-unit-things. The problem I have trying to speculatively compare that against Bulldozer's 480sp wonderkind is that you're comparing a low-end mainstream chip that will be out in late Q4 2010 to a high-end chip that won't be out until sometime in 2012 (in my opinion - I know there's no hard date for Bulldozer). There is enough trouble trying to compare SPs from differing architectures - we need to keep the market segment and timeframe at least somewhat in line if we're to make any reasonable comparison. If you were to guess how many SPs will be on a high-end Ivy Bridge when Bulldozer arrives, I would presume you wouldn't guess "6."

First of all there will not be a Bulldozer CPU with an integrated GPU in 2011.There will be a low power Bobcat core with low end EvergreenDX11 GPU(80SP,the level of what AT tested in SB) and a mainstream desktop/notebook chip with improved 10h/Phenom II cores (Quad/Dual) done on 32nm and with a 480SP Evergreen DX11 GPU. The last one is called Llano and this will go against SB with its "better" version GPU in mainstream desktop segment. Bulldozer is high performance Quad/Octal with no GPU on chip,at least not in 2011.In 2012 there will probably show up a whole line-up with on board Evergreen GPUs.

[JumpingJack]

Hi,

I would just like to chime in here and speak to some of the comparisons I have been seeing with regards to Sandy Bridge's IGP performance and Bulldozer's 480sp IGP. ."

Bulldozer will not have a 480sp IGP because it won't have an IGP.

This verison of SB will be up against Llano.

[natty]

Only 10% performance improvement over the 2 year old Nehalem (clock for clock)

Yawn.

[ajaidev]

BTW Why is there no i7 930,940 & 950 in the picture?

[slaveondope]

BTW Why is there no i7 930,940 & 950 in the picture?

They are in a different segment than this Sandy Bridge

[Hornet331]

Only 10% performance improvement over the 2 year old Nehalem (clock for clock)

Yawn.

lol better get your facts right before yawning

BTW Why is there no i7 930,940 & 950 in the picture?

Because theres no 2500 or 2600 to test?

If you want to compare it, just look it up in the cpu database.

[ajaidev]

They are in a different segment than this Sandy Bridge

Because theres no 2500 or 2600 to test?

If you want to compare it, just look it up in the cpu database.

But there is a performance difference between i7 9xx and the i7 8xx on the same clock. I am 1000% sure that the sandy bridge i5 2600 is not super fast compared to i5 2400 "c2c" but the cache difference is there...

[terrace215]

How about we wait and see them ... before making ... prediction

Because that would not be a prediction.

Look, given that SB turns out to be another leap, and not a saaya-like "1-2% gain on non-AVX", and now that we've had more BD design info disclosed, I think it is clear enough that SB will be in another league (than BD, let alone Llano) on client workloads. I guess you'll have to wait until Q4 2011 to verify (or not) this prediction, unless AMD makes client samples available before launch.

[informal]

Ok Nostradamus

[slaveondope]

Ok Nostradamus

Found Terrace an avatar

Jokes people jokes

[RaV[666]]

And for client, then consider that by the time BD launches for it, Intel will likely be shipping the *mainstream* versions of Ivy Bridge at 22nm!

There`s some bridges you maye be interested in aquiring.
Ivy bridge is expected to hit in 2012.Bulldozer has been officially announced as 2011 product.So, no.

[informal]

Found Terrace an avatar

Jokes people jokes

Hehe,good one man

[Lightman]

Very nice preview by Anand with sweet looking upcoming CPU from Intel.
What got me thinking is the IGP part of tests. Why all of them are in lowest IQ 1024x768?
I know that IGP is only for causal gaming but even Anand himself says that some of titles could be played in medium/high setting and still maintaining around 30FPS.
Is this because Intel explicitly asked to test like that?
Don't get me wrong, SB IGP looks like first proper Intel GPU since i740 especially considering that this was 6 'shader' version running at unknown clock (most likely non-turbo) but I have a felling heavier shader setting would put it more often behind HD5450 in FPS.

I just wish Anand would made these 2-3 more IGP tests to show some IQ/perf scaling of upcoming Intel solution.

[JumpingJack]

Very nice preview by Anand with sweet looking upcoming CPU from Intel.
What got me thinking is the IGP part of tests. Why all of them are in lowest IQ 1024x768?.

Go observe any IGP related review... even nVidia's best IGP (or AMD's best IGP for that matter) cannot play contemporary games at high resoultions/quality settings -- plenty playable at the lower res (1024x768, 1280x1024 type res, but 1980 x 1200 -- forget it, no IGP can give a good gaming experience there). He probably tested there because that is where he tested all his other IGP solutions and wanted to make a comparision.

For example, here is the 890GX, all low res and low quality settings to get into the 'playable' range:
http://www.anandtech.com/show/2952/2
http://www.tomshardware.com/reviews/...on,2571-7.html

[MAS]

SuperPi 1M

Q6600 @2.5GHz - 20 sec
Sandy @2.5 GHz - 16.349 sec

4 years passed, cpus didn't change

[mstp2009]

SuperPi 1M

Q6600 @2.5GHz - 20 sec
Sandy @2.5 GHz - 16.349 sec

4 years passed, cpus didn't change

Conveniently left out the AMD chips, eh?

BTW, the above is an 18.3% performance improvement. I would not call that "the same".

[terrace215]

Ivy bridge is expected to hit in 2012.Bulldozer has been officially announced as 2011 product.So, no.

Intel is due to *ship* 22nm in Q4 2011, like they will ship 32nm SB this Q4. The first parts will be the mainstream 1155 SB successor, Ivy Bridge.

I predict you won't see client BD launched until Q4 2011.

So, I stand by what I wrote.

[RaV[666]]

Well, Intels roadmap clearly states Q1 2012.
AMDs roadmap states 2011.

In best case scenario bulldozer launches around end of H1 beginning H2, worst case, in Q4.
Intels Best case is january 2012, which means Q1 2012, worst case end of Q1.
So, youre incorrect.

If by "shipping" you understand the fact that it will start to go from manufacturing plant to channel (not actual availability)
The same applies to AMD.So if launch for bulldozer will be in Q4, actual "shipping" will occur somewhat sooner.
So again, your statement is false

[terrace215]

So again, your statement is true

Fixed.

My statement, as can clearly be seen above, is that BD client launch and IB client ship will both be in Q4 2011.

As your recent post shows, you agree.

[570091D]

[Johnny87au]

Wow.. intel realy cut the L3 latancy quite a bit, even if its not the real number.. 26 cycles that the program reports is quite impressive (compared to 35 from lynnfield).

And finally its confirmed theres no S1356... well quad channle here we come..

So their isnt a high end coming like everyone had rumoured??