Anand Sandy Bridge performance preview is up

[BD2003]

Exactly. Anyone buying a chip that high end that's even concerned in the slightest bit about gfx is going to go discrete. same with virtually anyone that wants a K chip for overclocking.

It's the low end where integrated graphics are going to matter, yet it doesn't seem like they're addressing that segment on the desktop. I'd even be willing to pay a slight premium.

 

[DrMrLordX]

My guess is that the 6 EU parts are the way they are to hit lower TDPs. Go up to 12 EUs, and power consumption goes up.

 

[BD2003]

But I thought the laptop parts were to have 12EUs? Naturally they're going to be low TDP. Surely it can't be that cut and dry.

 

[DrMrLordX]

It may be that the EUs for the laptop parts run at a lower clockspeed, but I couldn't tell you that for sure.

Either that or it's just Intel finding ways to give low-end buyers an incentive to stick a midrange CPU into an otherwise budget system to improve their overall profit margin.

 

[bryanW1995]

It looks like Sandy is about 10-15% improvement in IPC, it's a wonder how intel keep improving IPCs years after year while AMD is like stagnant all this time. Anyways, excellent chip. But I don't know if it's worth it for current 1156 owners to buy new board+cpu to upgrade for it, of course you can also probably clock it to like 4,5. still overall, don't think it's worth it for current 1156 owners who's got i7 8xx or i5 7xx that's clocked 3,5-4 already. also another factor is what intel will do with pricing on K series and this half unlocked business for lower end. Also I want to see how far AMD has come with BD, that will actually change a lot of things or .. not, 2011 will be an interesting year. 2010 is just too dull in cpu market.

P.S: where did Anand get permission to do a full review on this thing while no other site has published anything on it!? i hope they don't get sued. don't get me wrong I love it!! Are they going all out for the 13th anniversary celebration of the site?

kyle at hocp actually referred people here to read anand's preview, I think that he said it's more up anand's alley. from reading that, I don't know if it's that anand got extra special treatment or if most of the others just weren't interested.

 

[bryanW1995]

*sigh*

*snip*

In all honesty DO NOT GET a 1155 system until you have seen BullDozer.
Im being serious about this.

the server version of BD should only be a few months behind skt 1155 anyway, it shouldn't be too hard to hold out until then.

 

[IntelUser2000]

It may be that the EUs for the laptop parts run at a lower clockspeed, but I couldn't tell you that for sure.

Either that or it's just Intel finding ways to give low-end buyers an incentive to stick a midrange CPU into an otherwise budget system to improve their overall profit margin.

Read the new mobile Sandy Bridge article from the front page. 650/1300MHz is not far away from 850/1150MHz on desktop SKUs. I'm guessing the "effective" clock speed difference is only 15% at most.

Another thing. If they can put a 12EU on a laptop part, they should definitely put one on desktop as well. Why? Because desktop needs a faster graphics just like mobile does. It may be somewhat less relevant, But desktops are still significant amount of revenue. Plus why'd they cripple a part when they need everything to counter Llano?

 

[DrMrLordX]

Read the new mobile Sandy Bridge article from the front page. 650/1300MHz is not far away from 850/1150MHz on desktop SKUs. I'm guessing the "effective" clock speed difference is only 15% at most.

Okay, so there is a clockspeed difference. If the desktop EUs need to run at a higher voltage (or are specced for a higher voltage anyway) to run at those speeds, that would account for some of the TDP difference, but not all of it.

Perhaps the rest is made up by the lower clockspeed of the x86 cores in the budget parts. That's a lot of maybes though . . .

Another possible issue here is the very nature of the EU as well. If it operates anything like the Clarkdale IGP, then it will consume a non-trivial amount of computing resources in the x86 cores in order to deliver maximum performance. Part of the problem here may be that adding more EUs works the x86 cores harder (on average), increasing overall average heat output/power consumption without necessarily changing maximum power consumption of the entire chip by all that much.

Another thing. If they can put a 12EU on a laptop part, they should definitely put one on desktop as well. Why? Because desktop needs a faster graphics just like mobile does. It may be somewhat less relevant, But desktops are still significant amount of revenue. Plus why'd they cripple a part when they need everything to counter Llano?

It was not my initial inclination to accuse Intel of arrogance, but that may be what we see here. In other words:

1). Intel may not fear Llano at all
2). Intel figures they can goad budget users into ponying up for 12EU chips just to avoid being screwed by buying into a "budget" 6EU chip

It has been a complaint as of late by anti-Intel partisans that Intel has a habit of crippling features on low-end chips just to push people into buying their higher-end, higher-margin products. With Sandy Bridge being as it is, perhaps that complaint is (somewhat) valid. You can't overclock (much) with budget Sandy Bridge either, especially considering how turbo speeds will delimit how much overclocking will be possible. The i3 Sandy Bridge chips don't even have turbo. Will they overclock at all?

Wouldn't it be a sad state of affairs if an i3-530 could smoke the i3-2120? It's not hard to push the 530 to 4.6 ghz without crossing the dreaded 200 mhz BCLK barrier (well, relatively simple anyway).

 

[IntelUser2000]

If it operates anything like the Clarkdale IGP, then it will consume a non-trivial amount of computing resources in the x86 cores in order to deliver maximum performance.

This type of quote is brought often enough for me to be slightly annoying, because well... that's not really true.

Core 2 on G965

Scores are Hardware/Software VS respectively

3DMark01-4357/7607
3DMark05-896/1464
3DMark06-545/723

You can see they are definitely faster on software. The problem with the G965 was that 3DMark used software results, while real games used hardware results.

Core i5 661
3DMark01-12919/12007
3DMark05-4006/2245
3DMark06-2089/1716

On the latest GMA, the hardware is significantly faster than software. I've seen a single program that deviates from the result, and that's 3DMark Vantage. No other games show faster result on software VS than hardware VS on the latest GMA.

While they do offer the switch via the graphics control panel, unless you have the ULV versions clocking at 166MHz, its highly unlikely you'll encounter a situation where software is faster than hardware. Even the ULV chip can do 500MHz Turbo.

 

[DrMrLordX]

Fair enough then, we'll just assume that reliance on the x86 cores will not be a weakness for the Sandy Bridge EUs and leave it at that.

 

[kalniel]

What do you mean by that?
Intel generally publishes quite detailed roadmaps, and they document their architectures pretty well in the Intel optimization manuals.

Once the products are released, certainly. I'm not trying to suggest Intel are all furtive when it comes to architectural previews, as AMD used to be the same, but, perhaps as a result of needing to sow hope in a future product, AMD seem to have been a lot more willing to discuss architecture of future products these days (cf. both Intel and nVidia).

 

[Scali]

Once the products are released, certainly. I'm not trying to suggest Intel are all furtive when it comes to architectural previews, as AMD used to be the same, but, perhaps as a result of needing to sow hope in a future product, AMD seem to have been a lot more willing to discuss architecture of future products these days (cf. both Intel and nVidia).

Problem with AMD is that usually the final product isn't as good as what they've been talking about for the past 12+ months.
Intel shows the products first, so you know exactly what to expect. How they managed to deliver that performance is interesting, but I think it's more important to have confirmation of the performance first.
Bulldozer is probably going to be another big letdown for most people, after AMD has gotten their expectations up.

 

[kalniel]

Maybe AMD's problem is that other people are claiming that their products are better than they are. I don't see any firm performance indications yet from AMD concerning Bulldozer, save the magny cours comparison, so I don't see how it can be a big let down because it's not AMD who are getting expectations up.

 

[Scali]

People just love a good fairy tale.

 

[Munky]

Cant help but notice SB requires a new socket... and think back of all the s1366 peeps booing s1156 as a dead end.

 

[Edrick]

I wonder what the RAM multiplers will be like. He states that the P67 boards will have unlocked memory multiplers and the H67 will be locked at intel spec. I understand that part, but what I am unclear on is what these multiplers will be.

Fo so long, memory multiplers have been a factor of 133. (So 10X on a Lynnfield would be 1333mhz for example). Now that SB has a 100mhz BLCK, how can we keep the familiar RAM settings without having some strange multiplers?

10.66X = 1066mhz
13.33X = 1333mhz
16.00X = 1600mhz

I suppose they can do that, but it does not keep things in even ratio like 1:2 or 2:8. I guess I will have to wait to see some Bios settings from these new MBs.

 

[ilkhan]

I wonder what the RAM multiplers will be like. He states that the P67 boards will have unlocked memory multiplers and the H67 will be locked at intel spec. I understand that part, but what I am unclear on is what these multiplers will be.

Fo so long, memory multiplers have been a factor of 133. (So 10X on a Lynnfield would be 1333mhz for example). Now that SB has a 100mhz BLCK, how can we keep the familiar RAM settings without having some strange multiplers?

10.66X = 1066mhz
13.33X = 1333mhz
16.00X = 1600mhz

I suppose they can do that, but it does not keep things in even ratio like 1:2 or 2:8. I guess I will have to wait to see some Bios settings from these new MBs.

Good point. Hopefully PC3-1600 is supported, then the rest doesn't really matter. Be nice to see some 1.3v 1600/8 (or even 7) RAM out when sandy gets released.

 

[PlasmaBomb]

It's still not that impressive vs my current rig...

What is my optimal gain? 4.4ghz with a 0-25% clock for clock advantage?

Color me not that impressed, considering my rig is now 3 years old and was mid range back then...

erm... using your own figures that's somewhere between a 15.2% gain and a 44% gain... (depending on IPC improvement).

 

[Acanthus]

erm... using your own figures that's somewhere between a 15.2% gain and a 44% gain... (depending on IPC improvement).

That would be worth it, if the 44% wasnt in winrar and video encoding. For the things i'd use it for, id be looking at the low end of that % figure.

This would require a new overpriced intel mobo, $~325 CPU, and at least 8GB of DDR3...

And i don't even get a hex core at the price because intel decided to use up their real estate on the graphics chip that will be disabled by any end-user that plays video games. Oh wait, i think it can play the original unreal tournament at low settings.

 

[khon]

That would be worth it, if the 44% wasnt in winrar and video encoding. For the things i'd use it for, id be looking at the low end of that % figure.

This would require a new overpriced intel mobo, $~325 CPU, and at least 8GB of DDR3...

And i don't even get a hex core at the price because intel decided to use up their real estate on the graphics chip that will be disabled by any end-user that plays video games. Oh wait, i think it can play the original unreal tournament at low settings.

So what you've established is that you're not the target audience.

 

[ilkhan]

So what you've established is that you're not the target audience.

No, we've established that what he wants != what he needs != what he is willing to pay for.

 

[kami]

I'm more interested in the LGA-2011 w/ quad channel memory for a true upgrade. Looks like my 15 month old i7 will be around for at least another 15 months. I have never used the same CPU for such a long time.

 

[IntelUser2000]

Update: it looks like we may have had a 12 EU part from the start.

Nice Anand, nice...

 

[Idontcare]

you can bet some llano designers just breathed a big sigh of relief

 

[khon]

you can bet some llano designers just breathed a big sigh of relief

I'm not so sure about that.

The performance SB showed is enough for all non-gaming purposes (Intel says it can do 3D 1080p decoding), and who buys integrated graphics for gaming ? Llano will be better, we all know that, but it still won't be enough for serious gaming, so to some degree you have to wonder, what exactly is the point ?