Sandy Bridge preview

Looks like the cpu performance are almost unchanged :|

It beats the similarly clocked i7 880 in most cases (sometimes HT plays a role), even though turbo is disabled, and even though it has less L3 cache (6MB vs 8MB).
 
Not sure why but the whole article seems like a damage control PR from Intel. This could mean two things:

a) Sandy Bridge is going to be late
or
b) Intel is trying to nullify the impact of AMD's upcoming processors.
 
With mature drivers and a fully functioning graphics turbo to add on top of that, its potentially a very capable little unit. Intel will definitely surpass the consoles with their next generation IGPs accross the board if they can keep up the good work. It should make for some killer laptops and SFF PCs.
What about DirectCompute and OpenCL support? If Intel can deliver drivers that make both of them a reality then it'll do absolute wonders for adoption.

I never thought I'd get so excited about Intel integrated graphics! :LOL:

Good point about comparing to current consoles. I'm sure Intel can provide support for OpenCL, but may choose to implement it using CPU rather than GPU resources. Since the CPU and GPU share the same memory subsystem from L3 and out, I can't really see that using the GPU would bring any benefit, whereas the L1, L2 cache hierarchy of the CPU for many codes can. So it will be for compatibility reasons at the behest of their customers if it happens. We'll see.

I'm impressed by what I see though - much more so than by Nehalem. That said, Bulldozer looks good on paper, and in consumer space it will clearly boil down to market positioning and OEM deals.

In my professional field, the quad channel memory of the higher end Sandy Bridges will definitely be a competitive advantage.
 
I would not consider "beating a tortoise" to be fast.

5400 is tortoise, old GMA is snail.

(5400 has 20 times less calculating speed than high-end chips)

So Sandy bridge, by beating 5400 with some 20% difference, is only 16x times slower than high-end chips!

It also uses something like 30 times less power than the highest end GPU on the market.

And Llano will have 3 or 6 times more shaders than 5400, so it should be somehting like 2-4 times faster than SB.

Llano is coming out in late 2011. It will have to compete with the Intel parts with turbo boosted CGPUs with twice the number of functional units tested here. Of course Llano will win against Sandy Bridge but it may not be so clean vs the 22nm refresh coming in early 2012.



With mature drivers and a fully functioning graphics turbo to add on top of that, its potentially a very capable little unit. Intel will definitely surpass the consoles with their next generation IGPs accross the board if they can keep up the good work. It should make for some killer laptops and SFF PCs.

It'll be interesting to watch. Who would have thought that an Intel IGP would even come close to a console this quickly. I bet you would never have predicted it 3-4 years ago! :p


I never thought I'd get so excited about Intel integrated graphics! :LOL:

Maybe you should pinch yourself to make sure you're not dreaming? :)

I never thought I would either! However this can only mean good things for the PC game industry if we can bring up the Intel range into something approaching respectability.
 
a) Why does it seem like damage control to you?

I don't remember seeing the "official" benchmark results of an upcoming processor so early before. Mix the benchmark with the AMD's Bulldozer articles from last week and with the Intel's lowered guidance for the third quarter and we have our soup ready :)

b) If it is damage control, then have done quite well, haven't they?

Yes, the results look really promising. It will be interesting to see how AMD's upcoming IGPs will compete with the Intel's future offerings.
 
IIRC, SB was said to have a DX10 GPU.

I am quite impressed by the elvel of performance they have been able to extract. Comparing it to Llano will be interesting.
 
Llano is coming out in late 2011. It will have to compete with the Intel parts with turbo boosted CGPUs with twice the number of functional units tested here. Of course Llano will win against Sandy Bridge but it may not be so clean vs the 22nm refresh coming in early 2012.

Llano shipments to OEMs were supposed to start Q4/2010, release and public availability on Q1/2011.
This schedule was delayed by about one quarter. So shipments will start Q1/2011, release and public availability Q2/2011. This is not late 2011, and this is about one year earlier than intel's 22nm.

I'd quess AMD propably has some bulldozer-based fusion out when 22nm intel chips ship on volume.
 
I don't remember seeing the "official" benchmark results of an upcoming processor so early before. Mix the benchmark with the AMD's Bulldozer articles from last week and with the Intel's lowered guidance for the third quarter and we have our soup ready :)

What damage exactly is there to be controlled? Isn't BD coming a fair bit later than SB?
 
What damage exactly is there to be controlled? Isn't BD coming a fair bit later than SB?

I think it could be stock damage control. The ambiguous, limited benchmarks plus the near zero architectural infos suggest that. They are starting to get hit by the affordable 6 core, and the relatively expensive Clarkdales. Combined that with Bulldozer and Bobcat infos that looks promising, investors might not want to keep the stock.
 
I never thought I would either! However this can only mean good things for the PC game industry if we can bring up the Intel range into something approaching respectability.

Well that's one way to look at it, another would be that a third graphics architecture to take into account makes the platform yet more splintered and costly for developers. And Intel finally having a modestly performing graphics chip is just half of the equation - the other being a solid driver set with feature support comparable to the other 2. We're not there yet by a long shot.
 
I think it could be stock damage control. The ambiguous, limited benchmarks plus the near zero architectural infos suggest that. They are starting to get hit by the affordable 6 core, and the relatively expensive Clarkdales. Combined that with Bulldozer and Bobcat infos that looks promising, investors might not want to keep the stock.

Huh? A few powerpoint slides from AMD won't move Intel's stock. Have you seen Intels' "actual" financial performance recently?
 
Holy damn...

The 'weak' IGP is effectively equal to the ATI 5450. The entire CPU package uses 10% less power than the i5/780 under load, but beats it by basically 20% or more across the board -- and more than a few cases it effectively equals or even exceeds the i7-980X. And not just in single-threaded theoretical stuff, either :)

And this is the $200 USD part? :eek:

And all that without ANY AVX optimization added into the software.

I hope the fix their Hardware Decoder as well since they are consistently using much higher CPU usage then other Hardware decode.
 
Anand just did a follow up focusing on the mobile variants, and this comment stood out for me:


Anand said:
The major difference between mobile Sandy Bridge and its desktop countpart is all mobile SB launch SKUs have two graphics cores (12 EUs), while only some desktop parts have 12 EUs (it looks like the high-end K SKUs will have it). The base GPU clock is lower but it can turbo up to 1.3GHz, higher than most desktop Sandy Bridge CPUs

http://www.anandtech.com/show/3876/intels-core-2011-mobile-roadmap-revealed-sandy-bridge-part-ii

Considering the desktop test was done with such early drivers on a 6 EU part, graphics turbo was said to be broken and Intel are supporting DDR3 1600 as standard, I'd say its pretty safe to double those IGP benchmark numbers for the mobile variants. That's insane, we're talking damn near console performance here. Can't wait till we see what the final hardware can do, laptop gaming just became about 10x more viable.

How are they packing so much goodness in a 35w envelope!
 
Last edited by a moderator:
Having world-beating manufacturing processes surely helps. I wonder if they'll ever release in-depth architectural details on the GPU though.

Oh, absolutely and I think the incredibly smart way turbo is managed is a big part of it as well. These chips deliver the power where your machine needs it in order to make the most out of those 35w, the whole idea behind an aggressive turbo mode split between the CPU and GPU just makes so much sense for mobile computers.

Yeah, I'd really like to see what they've done with their IGP, it surely can't just be trivial tweaks to get the sort of performance increase that they're getting.
 
The IGP component sounds real promising, providing a bigger leap in performance than previous generations. It may just satisfy the "average" users gaming needs.

I took a look at Anandtech's latest chart for mobile processors:
http://www.anandtech.com/show/3876/intels-core-2011-mobile-roadmap-revealed-sandy-bridge-part-ii

For higher demanding users, I want to say the 2720QM will be the "sweet spot", as I'm going to guess the price will be a good deal less than the 2820QM, and certainly much less than the 2920XM. The only difference of any significance between the two lower options is the 2MB of cache... I don't count 100Mhz much, even if turbo is engaged.

Will 2MB cache be of any value, especially considering the likely price difference between the 2820 and 2720 is likely to be $100+?
 
Back
Top