I think the update will bring a single cpu for now, then update later in the year with a dual processor. Will have to wait and see though. My current 2.8 will be fine until 2010. By then, we will see the single cpu being 16 cores total.
Single CPU? Where did you get that idea from? I don't see any reason whatsoever why they would artificially limit it to one CPU and then later add a second... Also, you will not see 16-core Xeons for dual-processor workstations in 2010... The DP Xeon of 32nm "Sandy Bridge" (out in 2011) is supposed to be 6 cores like the Westmere version. Probably won't see a 16 core DP Xeon until 22nm "Haswell" or later. (~2013).
Talk to Intel. Tell them to release chips more often. The last one had a 518 day wait to be updated. Whose fault is that? Intel.
Intel has nothing to do with Apple retaining ancient graphics cards and not upgrading RAM, HDD, etc.
So it looks like when the Mac Pro does go to Nehalem EP processors, we could see a serious performance boost.
The Nehalem Xeon 5570 is posting SAP transaction numbers over twice as fast as the Harpertown Xeon 5450 at similar clock speeds. Now SAP database transactions are a
very different kettle of fish from desktop Mac OS applications, but it does seem to point to the Nehalem EP Xeons being some serious performance monsters and the improvements just might be a bit better then the ~10% that many figured we'd see. Makes me glad I waited and I look forward to see how a Nehalem-powered Mac Pro benchmarks out.
I think part of the issue is that the new Quickpath and IMC architecture benefits MP systems much more than single processor systems, so many people looking at Nehalem/ i7 vs Core 2 Quad benchmarks are going to get the wrong idea of just how much of an improvement this is going to be.
The change in the MP/quad-processor server market is going to be even more dramatic considering the current Tigerton/Dunnington platform is completely starved for memory bandwidth in many applications.
Think about it.. Going from four (Core 2-based) quad-core Xeons using a legacy FSB and shared motherboard memory controller to a platform with four (Nehalem-based) CPUs that are native Octo-core and in which each have a dedicated 4-channel memory controller, 4x 6.4GBps CPU-to-CPU Quickpath links, 24MB L3, and capable of 16-thread SMT. It is going to be absolutely mind-blowing, and sadly the latest 4P Opteron platform won't stand a chance.
When serious applications are finally rewritten to full take advantage of multithreading, multiple cores and the GPU then the full weight of a hardware revision will be felt. Until that time there are few (real world) applications or scenarios that can bring a early-2008 MP to it's knees.
Granted not everyone who buys a Mac Pro needs all the power, but there are plenty of applications that can benefit from faster processors in the Mac Pro. HD Video encoding/effects, 3D rendering, Audio/Video analysis, and the simulations and calculations run in a thousand different scientific fields...
ATI =
ATI Radeon HD 3870 (you can buy one
here)
For Real Video professionals =
AJA Video's Kona.
These were found with a simple search.
The ATI card is a gamer card, not a professional OpenGL workstation card. The other link appears to be a video capture/encoding card.
Same with me.
I don't know of any interim resolution between 1920x1200 and 2560x1600 (except for different aspect ratios).
Unfortunately, it appears we are stuck with 1920x1200 for displays smaller than 30". I'm been fighting this battle for years, and believe we are not going to see high-DPI displays until both apple and MS make their OS interfaces completely resolution indepedent.
Nehalem is a new chip compared to Penryn, while Westmere is simply a die shrink.
1) Why wait for Westmere - will it be that much faster than Nehalem?
2) The wait shouldn't be as long for a die shrink as for a new architecture - right?
3)If you really wanted to, couldn't you pop a Gulftown into a Gainestown Mac Pro?
1) DP Westmere will have 6 cores and will be ~20% faster clock for clock.
2) At this point, it probably depends on the market more than anything, but Intel claims that they will push ahead at full speed.
3) It depends on if they change the socket/platform which is unlikely in the Xeon/server market. This will most likely be possible...
