backtop


Print 65 comment(s) - last by saratoga.. on Apr 28 at 12:51 AM

June 6, 2006 didn't fit AMD's schedule

We just got official confirmation from motherboard and chipset manufacturers in Taiwan -- AMD has moved the official launch date of Athlon 64 DDR2 up two weeks to May 23, 2006. AMD roadmaps have previously put the AM2 launch at June 6, 2006 (during Computex 2006), but since motherboards and CPUs are already completed, the launch will be pushed up. AMD insiders tell us Conroe's launch date was also a factor in pushing the AM2 launch date up, though even we do not know the exact date Intel's Conroe will launch.

AMD's latest advisories claimed the following:
  • May 16, 2006: Global announcement of Energy Efficient Processor roadmap and pricing
  • May 23, 2006: Global announcement of Socket AM2 and new desktop product availability and pricing
  • May 31, 2006: Global announcement of AMD LIVE! desktop system availability


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

reverse hyperthreading?
By sadffffff on 4/19/2006 3:35:29 AM , Rating: 2
i wonder if reverse hyperthreading will make it into the release? hope so, it sounds interesting.

http://www.bit-tech.net/news/2006/04/17/amd_revers...




RE: reverse hyperthreading?
By Knish on 4/19/2006 3:37:51 AM , Rating: 2
It's BS.. I am not a computer engineer or anything, but from everything CE engineers have told me, AMD just has to increase the pipeline width to get the same results. i think there was a bad translation


RE: reverse hyperthreading?
By MDme on 4/19/2006 12:49:02 PM , Rating: 3
you have a point about widening the pipeline. but it is exactly this point which make R-HT (reverse-hyperthreading) not BS and in fact may be feasible. "IF" AMD can make the two 3-wide cores appear as a single 6-wide core then it may translate to improvements. although this analogy would make two 2.0 ghz 3-wide cores into one 2.0ghz 6-wide core and not a 4.0Ghz 3-wide core in the process.

although a 4.0 Ghz 3-wide core might sound tempting. :)


RE: reverse hyperthreading?
By saratoga on 4/19/2006 4:35:39 PM , Rating: 2
"but it is exactly this point which make R-HT (reverse-hyperthreading) not BS and in fact may be feasible. "

How can you say that? We don't even know what it is. That link says nothing at all about what it does or how it works.

" "IF" AMD can make the two 3-wide cores appear as a single 6-wide core then it may translate to improvements. although this analogy would make two 2.0 ghz 3-wide cores into one 2.0ghz 6-wide core and not a 4.0Ghz 3-wide core in the process. "

I doubt thats what Reverse HT would do, since that is complete horse shit and could not actually be built. At least not in the sense you're thinking. The best you could do would be to build a 6 issue core, and give it two way SMT, and then call it two 3 issue cores. Of course, thats not so much "reverse HT" as it is "actual SMT" :) And yes, I am a computer engineer.

My guess is the actual system is just some sort of speculative execution that runs ahead of the main thread to prime the cache. Sun looked into such a system a while back. Its doable, but wastes enough power that I doubt it'd be very attractive unless you had a very low power system that wasn't ramping in clock speed like you wanted (which might be the K8, but I kind of doubt it).


RE: reverse hyperthreading?
By MDme on 4/19/2006 5:18:44 PM , Rating: 2
"How can you say that? We don't even know what it is. That link says nothing at all about what it does or how it works."

That's true, we don't know what it does. There are only speculations about it. What is said about it though is that it tries to use two cores to emulate one core. That being said, it may be indeed speculative execution but it is also possible that it may be making 2 cores appear as one wider core.

"I doubt thats what Reverse HT would do, since that is complete horse shit and could not actually be built. At least not in the sense you're thinking. The best you could do would be to build a 6 issue core, and give it two way SMT, and then call it two 3 issue cores. Of course, thats not so much "reverse HT" as it is "actual SMT" :) And yes, I am a computer engineer."

you are entitled to your own opinion whether it can be built or not. Oh and Mr computer engineer, if they built ONE six-issue core how can they give it SMT if it only had ONE core to begin with. ??? They can build two 3 issue cores and add SMT to the TWO cores but that is almost no different from current dual core CPUs.


RE: reverse hyperthreading?
By niknot on 4/19/2006 6:39:09 PM , Rating: 5
SMT refers to presenting the functional units of a core as separate logical processors. This allocation of execution resources can be done concurrently to compensate for a lack of ILP in the code being scheduled or sequentially to hide stalls caused by dependencies. There is no contradiction in providing SMT to a single-core processor. That however wouldn't be "reverse hyperthreading," it would be SMT.

Making a parallel execution engine appear to the software being run as a sequential execution engine isn't in of itself novel. That is what superscalar cores do already. A lot of hardware is added to decode, re-order, and speculatively execute instructions in parallel and present state that appears to be modified sequentially. As everyone knows--and this is the motivation of SMT in the first place--in certain types of code there is little ILP that can be obtained and functional units are left underutilized. Working with a simulator and sample code an engineer can model the effects of issue-width on the number of cycles necessary for executing code. When performing this modeling the engineer will find that while the size of the circuitry grows quadratically with issue width, performance does not. For typical code, as the issue width tends toward infinity, the rate of improvement tens toward zero. Naively increasing the issue width thus won't solve any problems, because the code being executed doesn't contain sufficient ILP.

Taking two n-issue cores and naively adding the necessary scheduling and re-ordering hardware to make them appear as a single core will never provide the same performance of a 2n-issue core. You're reducing the space complexity of an actual 2n-issue device, at the cost of genuinely sharing the same resources. At the lowest-end you'll waste a lot of resources performing speculative execution that will be thrown away 97+% of the time. Somewhere in the middle you'll schedule work on both processors realizing that functional units on both are going to be underutilized because of limitations in scheduling and re-order possibilities caused by both cores sharing only some state. At the highest end you won't pretend to be a single processor but will instead present logical processors to the operating system where some of the functional units of one of the physical processors will be scheduled and the results re-ordered as if they belonged to the other physical processor. This way the functional units in the second processor are not wasted prematurely when there operating system has work for them to do. This is necessarily less optimal than SMT on a 2n-issue device, but the cost of producing the device is lower in design and manufacturing.

What AMD has in the future is a problem: their cores are probably going to trail Intel's by 15%, and they have only so many resources to devote to remaining competitive and designing future products. While enjoy a commanding performance and power dissipation lead for years, AMD has only enjoyed small increases in market share that could evaporate more rapidly than it was obtained if Intel suddenly starts just being better. They cannot financially afford that sort of problem.

What AMD has, is a multicore process that is working quite well for them. What they also have, is that a lot of the software people use isn't aggressively multithreaded. Even if they can't obtain 2x improvements in performance in these programs through scheduling and synchronizing work on two cores, perhaps they can obtain enough to remain competitive with Conroe on computationally-intensive tasks with a lot of ILP. If AMD's is doing anything but tossing feces on the wall in hopes of preventing post-IDF rumors from sabotaging their future, it's probably that.


RE: reverse hyperthreading?
By rpsgc on 4/19/2006 5:57:20 AM , Rating: 2
It's for K10 (or whatever it's called).


RE: reverse hyperthreading?
By slyczar on 4/19/2006 7:55:19 AM , Rating: 2
I believe K8L is what you were thinking of.


RE: reverse hyperthreading?
By Spoonbender on 4/19/2006 8:13:26 AM , Rating: 2
Sounds like nonsense to me. It's like trying to drive two cars at once to go faster, rather than just make a bigger, faster engine.

*If*, and this is a big if, AMD and Intel has struggled with it for a decade or more, they can extract more parallelism from x86-based programs than ~3 instructions per cycle, the cheap, efficient, easy and simple way of taking advantage of this is to make a wider core. Just like Intel is doing with Conroe, but even here, they're only widening it from 3 to 4, and they don't expect all 4 ports to be fully utilized even.

The difficult, inefficient, slow, expensive and wasteful method is to force two cores to execute the same thread "in parallel". So called "reverse hyperthreading".
Sounds like something an AMD fanboy came up with because everyone knows Intel = evil, so anything they do is bad, and so if AMD does the "reverse", they'll do good.
Just a shame it doesn't make sense.


RE: reverse hyperthreading?
By hstewarth on 4/19/2006 9:23:44 AM , Rating: 1
It does sound like something that a Fanboy would come up with. In a lot of ways Hyperthreading is the early form of Dual Core.

This reverse hyperthreading if possible sounds more like joining of ALU's between two cores as one.

In time the best solution for performance is for soft developers ( especially game developers ) to designed their code more multi-threaded. It sounds like Oblivion is such a game that is more multi-threaded.

On the AM2, I think the pre-views have be so-so so far and not very impressive. I thinking AMD desiring to launch it early because they know when the Conroe comes out things will seriously change. So if they can get some buyers before the Conroe, it would mean more money for them.

Just my opinion..


RE: reverse hyperthreading?
By ZmaxDP on 4/19/2006 7:16:12 PM , Rating: 2
Not to knock the criticism, but "drive[ing] two cars at once to go faster, rather than just make a bigger, faster engine" isn't that crazy. Ever heard of drafting? Also, there are some concept cars with multiple engines in one car.

According to your analogy, a car is a logical core. An engine is a physical core.

Dual Core = Two cars with one engine each.
Single Core = One car, one engine.
New Thing = One car, two engines (or the two cars, two engines welded together at the bumpers.)

Like someone noted earlier, if they can gain 15% speed increase (or 15% efficiency increase) by "reducing the air friction" and "reducing weight" then they have made up for Conroe's gains on single threaded apps. (Obviously, two engines one car is ideal here)

To me it seems as if the trick is going to be finding a way to make it run both ways depending on the application demands.


RE: reverse hyperthreading?
By Steve Guilliot on 4/25/2006 12:55:59 PM , Rating: 2
There's no point in trying to hyper-analyze an analogy that doesn't apply.


RE: reverse hyperthreading?
By Chadder007 on 4/19/2006 9:13:31 AM , Rating: 2
OH snap....that sounds pretty cool.


RE: reverse hyperthreading?
By crystal clear on 4/19/2006 10:57:34 AM , Rating: 3
Cottection its X86- a french website.
They do not give their sources to prove reliability of the news.Sounds just a theory thrown in to create some news.


RE: reverse hyperthreading?
By crystal clear on 4/19/2006 10:59:10 AM , Rating: 2
sorry typing error it should read Correction.


RE: reverse hyperthreading?
By daschneider on 4/19/2006 11:39:09 AM , Rating: 2
You can't keep throwing parallel resources at a single threaded program and expect to get performance improvements. Combining two cores into one logical core may improve performace slightly but probably no where near the 2x cost of the resources.

Read up on Amdahl's law for the details: http://en.wikipedia.org/wiki/Amdahl's_law


RE: reverse hyperthreading?
By glennpratt on 4/20/2006 6:05:51 PM , Rating: 2
Umm, you do realize, the cost is already paid. The extra core(s) are already there. This will do nothing for good multithreaded app... assuming it's even real.


"We shipped it on Saturday. Then on Sunday, we rested." -- Steve Jobs on the iPad launch

Related Articles
AMD's Next-gen Socket AM2 Revealed
February 6, 2006, 6:42 PM













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki