MIT's 100-core CPU Will Be Ready This Year

But will it play Battlefield 3?

pshhh if my quad core can handle battlefield 3 how do you expect a 100 core to handle it? there's just no possible way.

-_^

Give me less cores...more GHZ! I don't care what it takes. Bring back the GHZ wars!

oh, and thanks in advance.

While a good approach, it will not be mainstream. People have been asking Intel to do this for years with the original atom processor. Now that Medfield is out and can run at peek frequency at 650mw. Intel could pack 100 medfields onto a single die for a fraction of the wattage of TILE-Gx, and have full x86 compatibility at the same time. If serious demand for this kind of chip surfaced Intel's designs and processes would be far more desirable.

and no it won't run BF3, or windows.

will they sell it or make a consumer version of it for market?

You sure it's not AMD or they're not stealing AMD's CPU strategy ?
*sarcasm*

inb4 DURR CAN'T WAIT TO PUT THAT IN MY GAMING PC from a bunch of 14 year olds who don't know what's going on.

I know this because I was that 14 year old once.

i sorta want GHZ too. The dolphin emulator requires it and a boat load of applications run faster too with higher GHZ. Hold at 8 cores and push the GHZ envelope to around 6...then add more cores again. Rinse and repeat.

You do remember the whole Pentium 4 vs Athlon vs Core2 fiasco right?
higher clocks do no necessarily mean higher performance......

"...pitted the TILE-Gx against Intel and AMD's Xeon and Opteron server-processors"

How is that for a test? They don't say how many cores the other servers had, so we can't make a real comparison here. It's no good having 100 cores if you only get a 67% performance boost over an 8-core machine.

Moreover it would help quite a few people to have 100 cores processors in desktop (non-server) machines too. I would certainly use 100 cores; I do some brute-force work as well as heavy floating point operations. Heck, I could do with a 1000 core CPU. Bring it on.

"higher clocks do no necessarily mean higher performance......"

This. You take a P4 3.80 GHz and I'll take a Core i7-2960XM [2.70 GHz]. The latter DESTROYS the former despite the difference in clock speed...

I'll take $$$ put into architecture changes that yield 10x performance improvements over the same money going towards doubling clock speeds...

o.O looks like i can run a hundred counter strike 1.6 servers from my home with enough ram and internet..
ILL BE RICH!!!!
btw keep it up.. reduces business costs.. saves space in servers.. this is just good..

Wait, is Facebook reviews hardware? I guess Apple fanbois found their safe haven with their $8k FB machines.

I wonder how F@H would run on this? 25x4 clients? 100x1 client?

x denoting cores fwi

Maybe not a review, but I assume they have some unbelievably huge amount of hardware powering their site, so I bet they have a vested interest in finding more efficient CPUs...

Sorry-- I replied to the wrong post

Outside of the fact that it's not an x86 processor, and thus not supported, the Wired article says that the cores are not very optimized for floating-point performance, which is what FAH depends on. 64, or 100, of these cores on one chip could make up for the individual lack of FP computational power, but I'm not sure any of us could say more than that.

blast from the past comment..., but will it run crisys?

i don't understand, don't GPUs now have 2048 cores
for stuff that needs parallelism GPUs are king, why use 100 core CPUs while you can get 2048 core GPUs

or is there something i missed

I think they should make a 100-core CPU with integrated graphics on a single chip. Working in 64-bit environment with a maximum TDP of 100 Watts, and supports DirectX 11. The integrated graphics have performance equivalent 4-way CrossFireX HD 7970. Too bad they have not think about it ...

actually i remember something on THW that showed all intel core processors had nearly equal IPCs
and they beat P4
so higher clock speeds on core processors do mean performance gains

That's called all the people donating the computers time for F@H.





Well, there isn't a core/thread limit as of right now and if memory serves right, on the F@H forums, someone had eight, 8 core xeons on one motherboard (64 cores but 128 Threads) and it was running fine with one client.

So 1 client will support over 100 cores. Although the better question is, is the FPU's in each core strong or are the weak? That'll make more of a difference.

Dear lord I hope you're joking. Nobody could be that clueless.

If it were an x86 chip why wouldn't it run windows? No reason not to run an enterprise grade server os on an x86 chip. Unless you are trying to save on licensing costs. But, I do agree that it would not be a great gaming chip. Same reason AMD got slammed on it's new architecture. I am pretty sure IPC on the atom chips wouldn't scale well for single threaded apps. But 100 atom cores might perform like 50 i7 cores. And not to mention the power savings if a serer can sit at idle running just a couple atom cores and throw on the heat when demand increases. I could see extra servers sitting around that can handle insane peak loads, but cost very little to operate when unused.

Using current silicon technology, I am pretty sure a 7970 won't run full load at 100 watts TDP, let alone the extra three, and add in the 100 cpu cores and you are sunk. You are describing like a 500 watt design. Now imagine what temperature that much energy in that small amount of space would reach. There is no solder invented on this planet that wouldn't melt. You'de have to bathe it in liquid nitrogen just to get it to boot.

You need to bone up on your CPU history, as do all the other people clamoring for "more GHZ."

The clock speed wars ended because it was totally unsustainable. The increase in heat generated from running at higher clock speeds is greater than the amount of extra speed you get. So, for example, if you increase the clock speed 15%, you increase the heat by 30% (not real numbers). Most people don't have access to plentiful liquid nitrogen, so you can't just keep ratcheting up the speed or the chips will literally burn up.

Instead, Intel and AMD have adopted different strategies to make the chips faster. Intel has aggressively optimized the chips to get a higher IPC (Instructions Per Cycle). This means that more gets done on every cycle. AMD was initially pursuing this as well, and was beating Intel in this arena, but in recent years they stagnated and even regressed (I'm looking at you Bulldozer) on their IPC. To compensate, they just cram more cores onto the chip. Works well in some cases, really poorly in others.

So, the chips we see today are a product of these diverging strategies solving the problem of "How do we make chips faster when we can't increase the clock speed?" Intel chips have fewer cores than AMD chips, but each core is much more powerful than an AMD core. This is why Intel totally dominates AMD in single threaded workloads. At a given clock speed, Intel chips are simply faster clock for clock than AMD's. On the other hand, AMD chips can really stretch their legs when presented a multithreaded workload and usually pull ahead of Intel if they can keep all their cores active.

GPU cores and CPU cores are not the same. GPU cores are actually extremely limited in what they can do, and what they're good at. If your application falls into the narrow bounds of what GPGPU is good for, then using a GPU is great. Otherwise, you want lots of CPU cores which are more general and more powerful.

so you mean that those 100 CPU cores aren't limited

If you want a many core, floating point optimized chip, look into Intel's upcoming Knight's Corner coprocessor. It's slated for a release either later this year or early 2013. I don't think they've actually said how many cores will be in it, but I've seen the number 50+ several times. It's also x86.

They demoed an early version several months ago, and the one chip was as powerful as the world's most powerful supercomputer from the late 90s. Now THAT would tear up some F@H.

I don't know anything about these chips, but no, they're probably general purpose cores. GPUs really only excel about performing the same operation on lots of pieces of data at the same time. So they're great for things like bitcoin, when all you need to do is calculate a hash on a ton of data, but lousy at things like running an operating system. Your computer is doing a lot of totally different things at once, and in fact most of the threads aren't even remotely related or similar. GPUs are terrible at that.

if you can put 100 cores on one die, then it's impossible they're not limited
if they weren't then why aren't they used now
i mean 64 core versions

usually i'd prefer having a quite good CPU for stuff that can't be GPU accelerated and a good GPU for stuff that works good on GPGPUs, like tirinity