• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Sandy Bridge architecture overview

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Acanthus said:
For me it will be all about where performance actually lands. I don't pay more than $300 for a CPU or more than $200 for a motherboard.
Click to expand...

Well it looks like a 2600k will be yours . If thats what your looking for. You may have to settle for the 2500k in your price range.
 
Shmee said:
not bad actually. A few concerns though:
-Hexcores, will they be available?
-Overclocking, and memory controller OC, as well as triple channel, are these supported?
-number of PCIE lanes? Will it be good for multi GPU setups?
-How will perfomance compare to a an Oced 980x?

All the above seem skeptical to me. Also, will the s1155 mobos be any good?
Click to expand...

No Hexacores on 1155 yet, LGA 2011 will be the performance platform.

Overclocking can best be described here: http://www.anandtech.com/show/3922/intels-sandy-bridge-architecture-exposed/8 Its possible that motherboards implement asynchronous data channels, however unlikely and most likely cost prohibitive for all except the most extreme overclockers.

SB-MB's System Agent has 16 PCI-E channels and another few (number?) coming off the Southbridge. It will support x8/x8 for multi-GPU

And for preliminary performance data:
http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row/1
 
so it seems the IGP was improved by
32nm process (westmere transistors)
L3 Cache partitioning
fixed function hardware units for everything bar the shading
beefed up shaders that only need to do shading

so while it may be possible to do DX11 as the shaders are significantly improved, you really, really don't want to (perhaps the drivers won't even support it), as that would presumably negate the benefits of the fixed function hardware units (Vertex processing, Rasterisation/Z, Texture Sampler, Pixel Ops + whatever Media Units which also attach to the Shaders).

this really reminds me of PS3 , Xbox360 type graphics being implemented in the IGP, not PC DX11 graphics. (although they did seem to replicate API ISA with CISC instruction on a 1 for 1 basis)
https://intel.wingateweb.com/us10/scheduler/catalog/catalog.jsp
 
Last edited:
How many recall when SSEII to came out . As programs took advantage IPC just kept increasing. The same exact thing applies here as AVX programms increase in number SB IPC will be magic and increase over its entire lifetime. So IPC of 20+ right now will keep growing . Another fact lost to most here is this . ALL SSEII programs are easy to recompile a simple prefix of vex is all thats required , We won't see the kind of performance increase as we will for NEW programms done with AVX inmind but we will see a good performance boost. AMD doesn't have prefix of Vex. That is intel exclusive.
 
the AVX seems more powerful than i expected, it doesn't have FMA (fused multiply and add) but can concurrently perform a multiply and an add at the same time.
 
Last edited:
dorion said:
I'm still waiting for Intel's reason that the socket had to be changed. I don't see anything from Anand's review that would require a new socket unless the 'Extensive power and thermal management for PCI Express and DDR' is more than just on die improvements.
Click to expand...
They integrated the clock generator for one. Most likely other reasons.
 
clock gen and other powersaving attributes seem more than enough reason to me to require a new socket.
and i see 1155 being short lived as tightly coupling flash memory would also justify a new socket
 
Last edited:
Am I the only one that is more interested on SB for laptops than desktops? Laptops certainly can use more CPU power, and finally a decent integrated GPU not be to avoided like the plague...whereas desktop CPUs once you OC are pretty much fast enough for most things anyway.
 
StrangerGuy said:
Am I the only one that is more interested on SB for laptops than desktops? Laptops certainly can use more CPU power, and finally a decent integrated GPU not be to avoided like the plague...whereas desktop CPUs once you OC are pretty much fast enough for most things anyway.
Click to expand...

Nope. I have been waiting to buy a new laptop for a year now. Skipped the 'dales completely. SB for laptops will be a well needed boost for that platform. As soon as HP has one for sale, I will be ordering.
 
There is indeed no DX11 support but DX10.1 isn't out of reach. They might even finally get anti-aliasing support, but we'll see.
 
StrangerGuy said:
Am I the only one that is more interested on SB for laptops than desktops? Laptops certainly can use more CPU power, and finally a decent integrated GPU not be to avoided like the plague...whereas desktop CPUs once you OC are pretty much fast enough for most things anyway.
Click to expand...
I went with westmere for a new laptop (last week even). Shrinks for laptops, new archs for desktop. At least for now. I'll pay attention to Ivy for my next laptop.
 
Edrick said:
So if I read the article right, thats equivalent to a 6K L0 cache? What type of performance increase do you see this having? Do you see this as being the standard doing forward?
Click to expand...

L0? Would registers be referred to as L0 as well, if we are going to go with an L0 moniker?

So we have an L0 I$ (uops) and an L0 D$ (registers) now?
 
ilkhan said:
I went with westmere for a new laptop (last week even). Shrinks for laptops, new archs for desktop. At least for now. I'll pay attention to Ivy for my next laptop.
Click to expand...

Are we thinking that basically SB laptop cpu SKUs and Llano laptop SKUs will hit the markets more or less at the same time?

There is always a longer delay for laptops before you can actually buy them, when are we realistically thinking SB laptops are going to be shipping to DELL customers?
 
I wouldn't put the register in the same boat as the new uop cache. The uop cache is really a cache while registers are specific to instructions.

But yea, "L0" is just a moniker.
 
Idontcare said:
Are we thinking that basically SB laptop cpu SKUs and Llano laptop SKUs will hit the markets more or less at the same time?

There is always a longer delay for laptops before you can actually buy them, when are we realistically thinking SB laptops are going to be shipping to DELL customers?
Click to expand...
The models will get the upgrade at different times. Consumer laptops first, business models as they finish testing.

Is llano bobcat or phenom based for the CPU portion?
 
StrangerGuy said:
Am I the only one that is more interested on SB for laptops than desktops? Laptops certainly can use more CPU power, and finally a decent integrated GPU not be to avoided like the plague...whereas desktop CPUs once you OC are pretty much fast enough for most things anyway.
Click to expand...

Yeah, the improvements are definitely bigger for laptops than for desktops, but it'll be nice for desktops as well. But this is just smart engineering and marketing because the laptop market is bigger than the desktop market at this point, right? 🙂 So as far as I can see, Intel devoting more resources to making their chip even better for laptops is doing what they should be doing to make their investors happy. 🙂 And they really went all-out with power management in SB, that's for sure. And it's sure not like it's bad for desktops, it's going to be better for them too, just not as much better. =)

But yeah, something huge like "oh, the new sb laptops have 30 mins (or an hour or whatever)" better battery life than our previous ones is something huge and tangible to market, which is really nice.
 
Saw one thing from IDF that I don't quite understand. Gigabyte was there and showing off their upcoming P67 boards, one of which looks like this:

p67ud7.jpg


That an awful lot of PCIe slots for a chipset that's only supposed to support 16 lanes for graphics, so what is up ?
 
I noticed that too. Also it seems to have 4 USB 3.0 ports. I'd say it is because the P67 chipset has full PCIe 2.0 support (5 GT/s), where as the P55 chipset was limited to 2.5GT/s. Anand mentioned it in the Sandy Bridge preview article:

The other major (and welcome) change is the move to PCIe 2.0 lanes running at 5GT/s. Currently, Intel chipsets support PCIe 2.0 but they only run at 2.5GT/s, which limits them to a maximum of 250MB/s per direction per lane. This is a problem with high bandwidth USB 3.0 and 6Gbps SATA interfaces connected over PCIe x1 slots. With the move to 5GT/s, Intel is at feature parity with AMD’s chipsets and more importantly the bandwidth limits are a lot higher. A single PCIe x1 slot on a P67 motherboard can support up to 500MB/s of bandwidth in each direction (1GB/s bidirectional bandwidth).
Click to expand...

http://www.anandtech.com/show/3871/the-sandy-bridge-preview-three-wins-in-a-row/3
 
More graphics info on Sandy Bridge.

-Supports up to 4 displays independent or concurrently
-4x Multi-sampled AA(the earlier parts support 1x, so nothing. 😛 )
-With AA they also added DX10.1 support
-OpenGL 3.0

Some CPU info I forgot to mention.

-SHA acceleration(bye bye Via)
-faster ADC and arithmetic multiply
 
Last edited:
khon said:
Saw one thing from IDF that I don't quite understand. Gigabyte was there and showing off their upcoming P67 boards, one of which looks like this:

p67ud7.jpg


That an awful lot of PCIe slots for a chipset that's only supposed to support 16 lanes for graphics, so what is up ?
Click to expand...

Looks like it has NV 200 chip
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top