First Sandy Bridge Numbers?

[lopri]

http://techreport.com/discussions.x/19061

Considering the source ("Coolaler"), I have to believe this thing is real. Numbers are likely very premature but there look to be some hints of what's to come. I've saved a couple of pictures for the lazy.

 

[lopri]

BTW the first Everest bench is called "CPU Photoworxx", and according to Lavalys it does this:

CPU PhotoWorxx

This integer benchmark peforms different common tasks used during digital photo processing.

It performs the following tasks on a very large RGB image:

· Fill
· Flip
· Rotate90R (rotate 90 degrees CW)
· Rotate90L (rotate 90 degrees CCW)
· Random (fill the image with random coloured pixels)
· RGB2BW (colour to black & white conversion)
· Difference
· Crop [EVEREST Version 2.10 and later]

This benchmark stresses the integer arithmetic and multiplication execution units of the CPU and also the memory subsystem. Due to the fact that this test performs high memory read/write traffic, it cannot effectively scale in situations where more than 2 processing threads used. For example, on a 8-way Pentium III Xeon system the 8 processing threads will be "fighting" over the memory, creating a serious bottleneck that would lead to as low scores as a 2-way or 4-way similar processor based system could achieve.

CPU PhotoWorxx test uses only the basic x86 instructions, and it is HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.

 

[Dekasa]

Hmmmm, I sure hope it does better than that. Seems to me that it's "just" on par with i7, clock for clock. Almost seems like something is wrong, personally.

*Edit*

After viewing the Everest benchies, maybe it's just suffering from a few bugs (either in SB or in the motherboard) or maybe I read about video card driver issues too much.

 

[Idontcare]

Originating thread: http://forum.coolaler.com/showthread.php?t=240578

Totally laughing my ass off at the fact that these benches and thread infos from coolaler are apparently too xtreme for XS forums...they keep intentionally deleting his thread and anyone else who tries to create a new one in regards to this info.

Here's what I don't get...it appears to me that Coolaler is doing all the gpu benching using a 5870 instead of the native on-die GPU. Why? Wouldn't we be more interested in seeing how good or bad Sandy's IGP is versus how well a 5870 benches on a Sandy platform?

 

[IntelUser2000]

Here's what I don't get...it appears to me that Coolaler is doing all the gpu benching using a 5870 instead of the native on-die GPU. Why? Wouldn't we be more interested in seeing how good or bad Sandy's IGP is versus how well a 5870 benches on a Sandy platform?

I wouldn't be surprised because its "integrated graphics". After all, its XtremeSystems guys we are talking about here.

There was a leak of 3DMark2001 benchmark for the GMA HD on the Clarkdale which showed a score of ~1100. That's less than 1/10th of what the production version could do. Who wants to bench THAT?

 

[Acanthus]

AMD has an opportunity if those numbers are accurate.

 

[lopri]

Well, I don't know whether turbo was at work but 16 secs for 1M @2.5 GHz looks pretty amazing?

 

[khon]

Maybe I'm missing something but those numbers look nice to me. At 2.5GHz it's beating out a 3.33GHz i7, which makes it about 35% faster at a given frequency. And the info I've seen has it running at similar or slightly faster frequencies (up to 3.8Ghz ?) than the core i7, so maybe 35-40% faster than the fastest current quadcore.
On top of that you get the reduced power consumption, and an integrated graphics core, sounds good to me.

Also it's worth keeping in mind that the quadcore sandy bridge is the mainstream product, the high-end stuff is 6-8 cores.

 

[alyarb]

Integer performance has certainly improved with this arch, it sucks P/H55 users can't do a drop-in upgrade.

By comparison the high end is looking even more luxurious than Bloomfield. Because of the 32-lane on-die controller and quad-channel DDR3, the size of Patsburg's socket puts it in an alarmingly elite and expensive context. Hopefully everything will slot right into the Bloomfield price structure but I doubt it.

 

[BTRY B 529th FA BN]

The socket is what I am thinking about. I just can't believe a 2011 socket will be out in 2011 -

I guess I'll believe it when I see it.

 

[alyarb]

wow I didn't even notice that heh.

 

[extra]

So per clock it looks like (from all the discussions i'm reading on it) about a 10% improvement per clock. And supposedly will be clocked a bit higher. So probably around a 20% improvement in real world performance. Which is about what most of us predicted. 20% improvement over Nehalem in a single threaded app is NOT a bad thing, folks! Looks like higher on some synthetic benchmarks, but meh hehe.

Woot. I am planning on building a Sandy Bridge system next year for sure. Well, what can I say...I live in Oregon...*right* by the Sandy river actually. It'd just be wrong of me not to buy one. . .

And yes of course AMD has a chance, why wouldn't they? Yeah there was that one thread where someone was saying "oh, sandy bridge will be like 100% faster blah blah" but the sane people knew that was nuttttters. Just like the ones saying bulldozer is going to be so good it'll make it look like the K8 days again...Nah, doubtful. I expect bulldozer to be a bigger improvement, % wise, over phenom, than sandy is over nehalem though. I plan on building a bulldozer system as well, I am super excited about it.

 

[IntelUser2000]

That's ridiculous. 4-channel DDR is the new EX platform.
http://pc.watch.impress.co.jp/img/pcw/docs/364/200/html/04.jpg.html

One site wrongly claims 4-channel on the high-end and every other site does copy+paste.

16.6 seconds for 2.5GHz is actually slower than current Nehalem.

 

[Ben90]

Also take into account that I doubt sandy bridge doesn't have any turbo profiles setup so its running at a true 2.5 while the other processors may be turboing

 

[IntelUser2000]

That's not a big deal. The advantage of Turbo Mode for other processors in multi-thread is only 4%.

The early Bloomfields without Turbo did better than this on SuperPI.

 

[taltamir]

Here's what I don't get...it appears to me that Coolaler is doing all the gpu benching using a 5870 instead of the native on-die GPU. Why? Wouldn't we be more interested in seeing how good or bad Sandy's IGP is versus how well a 5870 benches on a Sandy platform?

Not really... it will be "fast enough" for non gaming use, and gamers care more how it works with a real video card... like the 5870

Although I admit I am curious...

 

[Idontcare]

Not really... it will be "fast enough" for non gaming use, and gamers care more how it works with a real video card... like the 5870

Although I admit I am curious...

Why would a gamer buy sandybridge? That's not what the SKU is designed for.

This is like someone benching Llano or Bobcat but deciding to use SLI-GTX480 instead of benching the Fusion aspect of the CPU because hey there are some gamers out there that might want to see such benches...but that's not an excuse to not bench the part of the chip that defines its very existence.

Even if the performance sucks, from bad driver or bad bios support, its not coolaler job (or is it?) to make sure Intel's prototypes are seen in the best possible light so you'd think a leaker would leak the results as they are. It kinda defeats the purpose of leaking stuff if you only leak the stuff that is pretty much irrelevant to begin with.

 

[IntelUser2000]

Why would a gamer buy sandybridge? That's not what the SKU is designed for.

Isn't that kinda different? I mean what if we replace "Llano" and "Bobcat" with Bulldozer+GPU.

It's not like the upcoming 6 and the 8 core Sandy Bridge chips will help gaming performance anyway. And 3-channel memory? Doesn't even do much for Nehalem.

I think the single high-end GPU users could buy this version.


On a side note: Apparently there's both H65/H67 and P65/P67 chipsets. It's hinting that there are LGA1155 Sandy Bridge chips that lack GPU functionality. There was an early rumor which suggested 6 cores might be possible for the LGA1155 socket.

 

[Nemesis 1]

So per clock it looks like (from all the discussions i'm reading on it) about a 10% improvement per clock. And supposedly will be clocked a bit higher. So probably around a 20% improvement in real world performance. Which is about what most of us predicted. 20% improvement over Nehalem in a single threaded app is NOT a bad thing, folks! Looks like higher on some synthetic benchmarks, but meh hehe.

Woot. I am planning on building a Sandy Bridge system next year for sure. Well, what can I say...I live in Oregon...*right* by the Sandy river actually. It'd just be wrong of me not to buy one. . .

And yes of course AMD has a chance, why wouldn't they? Yeah there was that one thread where someone was saying "oh, sandy bridge will be like 100% faster blah blah" but the sane people knew that was nuttttters. Just like the ones saying bulldozer is going to be so good it'll make it look like the K8 days again...Nah, doubtful. I expect bulldozer to be a bigger improvement, % wise, over phenom, than sandy is over nehalem though. I plan on building a bulldozer system as well, I am super excited about it.

Hay you want to link that topic your talking about . I have the link . But your the one shooting off mouth. Lets see the link.

 

[taltamir]

1. More and more game are multi core capable. Especially the intensive games.
2. Sandy Bridge should still be faster than nehalem in general computing, and in games.

 

[Nemesis 1]

Why would a gamer buy sandybridge? That's not what the SKU is designed for.

This is like someone benching Llano or Bobcat but deciding to use SLI-GTX480 instead of benching the Fusion aspect of the CPU because hey there are some gamers out there that might want to see such benches...but that's not an excuse to not bench the part of the chip that defines its very existence.

Even if the performance sucks, from bad driver or bad bios support, its not coolaler job (or is it?) to make sure Intel's prototypes are seen in the best possible light so you'd think a leaker would leak the results as they are. It kinda defeats the purpose of leaking stuff if you only leak the stuff that is pretty much irrelevant to begin with.

Ha when the topic went up at XS I wondered the same thing. Than Who stepped in and put a stop to the topic . As you know their is that pay channel at XS. Ya might check that out.

 

[khon]

Why would a gamer buy sandybridge?

Is that a trick question ? Because it'll most likely be the fastest CPU on the market.....

 

[formulav8]

Khon, he could mean in reference to using the ondie gpu. Which he is exactly right. The GPU will stink overall for anything above low-end gaming range.

AMD's fusion is the only chip it seems that 'might' provide decent mid-range gaming. This is where AMD will have Intel beat quite well. Unless Intel pulls out a rabbit or something. Or AMD's gpu performs much less than expected.

One thing for sure is if AMD can get decent software support for Fusion then they will be ahead of Intel in quite a few area's. But we''ve already seen AMD trying to get software support in the past and made things like 3DNow! nearly useless for a long time. Even AMD64 has just gotten popular within the last couple years even though it was released many years ago.

For AMD, software support is key to compete with Intel in many non-gaming computing area's.


Jason

 

[faxon]

Why would a gamer buy sandybridge? That's not what the SKU is designed for.

This is like someone benching Llano or Bobcat but deciding to use SLI-GTX480 instead of benching the Fusion aspect of the CPU because hey there are some gamers out there that might want to see such benches...but that's not an excuse to not bench the part of the chip that defines its very existence.

Even if the performance sucks, from bad driver or bad bios support, its not coolaler job (or is it?) to make sure Intel's prototypes are seen in the best possible light so you'd think a leaker would leak the results as they are. It kinda defeats the purpose of leaking stuff if you only leak the stuff that is pretty much irrelevant to begin with.

perhaps because if it offers a significant enough single threaded boost in performance, there are a whole ton of notoriously cpu limited games (WOW, EQ2, ARMA2, SUPCOM2 just to name a couple) which gamers will want it for? we havent had a significant boost in single thread performance since conroe (penryn was mainly raw clock increase and a small boost due to cache size), and since nehalem was mainly aimed at working on intel's shortcomings in the server/workstation area, this next gen has to provide a boost in single threaded performance or else there's going to be a lot of people out there who arent going to be happy. there's still plenty of tasks which cant be written to support multicore processing (at least easily), and they have waited a long time now to be thrown a bone. AMD has caught up quite nicely in single thread performance in the mean time as well, and is offering amazing value with 6 core CPUs for $200. intel HAS to provide a single threaded performance boost this gen or else AMD is going to overtake them again, and if they do then AMD will become the gamers choice once more

 

[Idontcare]

If I dragged this horse any closer to the water I'd be drowning it.

I hear Bulldozer has a single-FPU unit per module...I sure hope the LAST thing anyone with a BD sample bothers to check is how the FPU performance stacks up. Why would anyone care to check out one of the key microarchitecture differences of a new chip? beats the shit out of me I guess