The Sandy Bridge Preview
by Anand Lal Shimpi on August 27, 2010 2:38 PM ESTThe Test
As was the case with Lynnfield, the current Sandy Bridge CPUs Intel is sampling are slightly different than what will be sold. The Core i5 2400 runs at 3.1GHz, has four cores, 6MB of L3 cache but no Hyper Threading. In order to help Intel’s partners test HT functionality however, the i5 2400s being sampled right now have Hyper Threading enabled. For the purposes of our test I’ve run with HT both enabled (to give you an idea of higher end SB parts) and disabled (to give you an idea of i5 2400 performance).
The other major difference between what’s out today and what’s coming in Q1 is turbo. Early Sandy Bridge samples, ours included, do not have turbo enabled. The CPU simply runs at 3.1GHz all the time, regardless of workload. The final retail 2400 will be able to run at up to 3.4GHz.
In other words, what we show here should be indicative of final performance, but it's probably slower than what will ship in Q1.
On the GPU side, the part I’m testing appears to be the single-core GPU configuration (6 EUs). Intel hasn’t released any info as to what parts will get the dual-core/12 EUs GPU configurations, although it may make sense for Intel to use the 12 EU parts in notebooks given the importance of integrated graphics to the mobile market. Update: The part we're looking at may actually have been a lower clocked 12 EU part, we're still waiting for additional confirmation.
Our test platform was a H67 based motherboard running with 4GB of DDR3-1333, the same memory we use in our Lynnfield testbeds.
I’m comparing to four other CPUs. The Core i7 980X for a high end comparison, the Core i7 880 for a near clock-for-clock comparison (albeit with HT enabled), the Core i5 760 for a potential price comparison and the Phenom II X6 1090T. The latter should be AMD’s fastest offering (if not close to it) when Sandy Bridge ships. Update: Note the Core i5 650 is actually the predecessor to the Core i5 2400, however I didn't feel a dual core vs. quad core comparison was too fair. The i5 760 will actually go head to head with the higher clocked i5 2500 when it launches in Q1.
| Motherboard: |
ASUS P7H57DV- EVO (Intel H57) Intel DP55KG (Intel P55) Intel DX58SO (Intel X58) Intel DX48BT2 (Intel X48) Gigabyte GA-MA790FX-UD5P (AMD 790FX) |
| Chipset Drivers: |
Intel 9.1.1.1015 (Intel) AMD Catalyst 8.12 |
| Hard Disk: | Intel X25-M SSD (80GB) |
| Memory: |
Corsair DDR3-1333 4 x 1GB (7-7-7-20) Corsair DDR3-1333 2 x 2GB (7-7-7-20) |
| Video Card: |
eVGA GeForce GTX 280 (Vista 64) ATI Radeon HD 5870 (Windows 7) |
| Video Drivers: |
ATI Catalyst 9.12 (Windows 7) NVIDIA ForceWare 180.43 (Vista64) NVIDIA ForceWare 178.24 (Vista32) |
| Desktop Resolution: | 1920 x 1200 |
| OS: |
Windows Vista Ultimate 32-bit (for SYSMark) Windows Vista Ultimate 64-bit Windows 7 x64 |
Facebook
Twitter
RSS
200 Comments
View All Comments
Anand Lal Shimpi - Friday, August 27, 2010 - link
We'll have to wait a little bit to find out... :)hnzw rui - Friday, August 27, 2010 - link
If it doubles Clarkdale's GPU performance, then it probably will (at least on lower resolutions). I'm getting pretty decent framerates from Clarkdale on 1360x768 Low and I've been able to play on 1360x768 Medium with a Radeon HD 4550. I think Sandy Bridge is probably closer to the latter than the former in performance.SteelCity1981 - Friday, August 27, 2010 - link
Any word when Intel will launch a mobile version of this new platform?Anand Lal Shimpi - Friday, August 27, 2010 - link
Q1 2011 :)BSMonitor - Friday, August 27, 2010 - link
So glad I waited on a 13" macbook pro! Sandy Bridge will probably be the next revision for Macbook's ehh?cheinonen - Friday, August 27, 2010 - link
Exactly my thoughts, that the GPU performance looks to be good enough that Apple could use it for the 13" MBP refresh next year. I'll be glad that I decided to wait, that's fur sure.synergist - Friday, August 27, 2010 - link
I was doing some research and if they would have to use the full integrated graphics core, with the 12 cores, to top the performance of the 320M in the current macbook pro 13 I doubt apple would take a step backwards in gfx performance, and use the 6 core integrated gfx.and the performance would still be pretty close, that the 320M would lose to the inter grated gfx (12 cores) by about 10-13%
and llano is still an option, but I have a feeling it will be a dead heat with this.
starfalcon - Friday, August 27, 2010 - link
Apple has to know a lot of the people buying the laptops are far from high end gamers.The amount of people with 320Ms who don't need them is probably a lot.
We'll see how all the different parts of Sandy Bridge work out.
Don't the Core iX processors not work with Nvidia Integrated graphics at all?
JarredWalton - Friday, August 27, 2010 - link
Correct on NVIDIA IGPs not working with Core 2010 (and presumably beyond). The need the QPI interface and Intel isn't licensing that to them.As for Apple, one thing to note is that they've started shipping all laptops with GPUs that support OpenCL it seems, so if Sandy Bridge doesn't have that they may do the same Optimus-style setup as current MBP. Not sure what they'd do with the base MacBook in that case, but Apple seems like they're gearing up to start leveraging OpenCL at some point in the near future. Pure speculation, though, so maybe SB's IGP will be enough, even if it's a step down from G320M.
DanNeely - Friday, August 27, 2010 - link
Aside from on the high end (LGA 1366/2011) the bus nVidia needs is DMI, not QPI. If I was nVidia I'd insist on getting rights to both because QPI is more futureproof. Specifically having more than a few high speed SATA6GB/USB3/etc devices will be able to saturate DMI since it's only the equivalent of a PCIe 4x slot (1.0 speed for 1156, 2.0 for 1155/2011) while QPI is a much higher capacity bus similar to AMD's HyperTransport.While intel seems determined to milk as much out of the (presumably) cheaper to implement DMI bus as it can; sooner or later they're going to either need to mainstream QPI or have the CPU die eat the SATA/USB3 controllers. I find the latter unlikely because it would require cramming even more data lines into the already overcrowded CPU socket area.