I guess this just goes to show how late AMD's 65nm process really is.

#2 and #49 I will just quote...

quote:
Will Half-Life 2 support Hyper-Threaded CPU's?

Gabe Newell: Hyper threaded CPUs attempt to extract thread-level parallelism, as opposed to traditional pipelined architectures which attempt to take advantage of instruction level parallelism. Hyper threading can be somewhat unpredictable in terms of the performance impact, as you can, in some cases, run slower.

Implementing and maintaining a "deeply" multi-threaded version of Source would be a pain (i.e. multi-threading the renderer). Implementing a hacky version (e.g. having a discreet physics thread or running the client and server in different threads) is something we may do depending upon how much bandwidth we have before we ship. Right now we don't get nearly as much bang for the buck working on hyper threading as we do on other optimizations. That may change as we run out of things to optimize.

64-bits, in contrast, is a one-time cost and is fairly simple to take advantage of. It's a huge win for tools as it not only gets more work done per instruction, but it also gets us past the current memory limitations, which are a problem for us today on tools.

Distributed computing is harder than hyper threading but it has the potential to increase performance by a huge amount (8X on our tools) as opposed to hyper threading (30%). All of our tools are going to a distributed approach.

So the taxonomy looks like this:

- general algorithmic optimization (general good thing to do)
- DX9 optimization (big gains, long term direction)
- 64-bits (not that hard, solves memory problem as well as performance gains)
- hyper threading (hard initial cost, on-going code maintenance cost, limited unpredictable performance gains, benefits in multiprocessor environments as well)
- distributed computing (hardest to do, biggest potential gains, great for tools, may be great for servers, not sure how it works with clients)

I predict that single core 32 bit CPU's will still be in the majority of PC's in 3 years......in 5 years they will still have a large market share.

Multi-core computing is a long way off for mainstream. Just look at 64bit computing.....how long have we had 64bit x86 CPU's....2 years....and yet there are no games and very few applications that are written for it. Even the next version of Windows will come in 32 bit and 64bit.

Right now is hype to get the hardware freaks to buy it.

now we need to see more software which supports multithreading so all this dually and quad action can be better taken advantage of, would be especially nice if games and every day apps would start to take advantage of that. If it's possible or even feasible

what use would it server outside of a cluster/ server farm?

I Guess Dual Core GPU's will be on the plate next for the Green and Red Boys. mmmmm...tasty.

I can see the use for quad-cores to the average desktop user. It just means they'll be able to have that many more programs/applications/spyware/viruses running at once. Heh, in 2007, you'll need a 21" wide monitor just to veiw the average users system tray.

Hmmm.... chipsets that contain L3 cache? On the chipset die I imagine. So then is that an HT link to L3? Is an HT link faster than the integrated memory controller? I mean the cache itself would be faster (SRAM on the chipset die vs. any kind of DRAM is no contest I would imagine) but what about the connection?

In 2007, will Opterons be using socket M2, or still 940? I wonder if anything would prevent the possibility of quadcore socket 940 chips..

Here we go, AMD is pushing this ahead a little in response to la Intella's more aggresive roadmap. Let the competition begin! We haven't had much of that for a long time :)

This is kind of funny -- I guess I had always assumed 2007 -- I hadn't even realized that AMD had previously said 2008.

Given that they have no new core coming out over the next two years, it seems that they almost have no choice but to go quad-core as soon as they possibly can (if they want to continue to take marketshare from Intel in the server space).

I hope they make an Opteron 1xx quad core, or an A64 X4, that isn't too expensive (i'd pay $500) -- I could definitely use it in my PC, but I'd rather not pay megabucks for it.