Reverse Hyperthreading does not exist @ TheInquirer

[werty316]

I guess the Core2Duo making AMD sweat is this.

Source (http://www.theinquirer.net/default.aspx?article=32885)
[qute]Apologies to the fanbois, but I am about to kill some nice AMD dreams. Reverse Hyperthreading does not exist. It's one of the bigger loads of bluff and bluster cooked up in a long time.

AMD doesn't have any such technology. It would like to have it but even the upcoming K8L quad core won't be able to present a dual or quad core as a single core to the Operating system.

On the other hand AMD doesn’t not need such a marchitecture as the operating systems are getting used to a dual and later quad machines. It's just a dream that someone set afloat but AMD doesn't have such a marchitecture.

However, AMD should weather the Conroe storm but it is about to lose the performance crown for a whole quarter at least. Its response, K8L won't be ready before Q2 2007. µ

* Apologies, too, for calling HyperThreading HyperTransport earlier. Must've been HyperVentilating. [/quote]

[Kougar]

Well at least the date for K8L is spot on target for the exact middle of 2007... That'll be worth far more than HT or RHT tech really. ;)

[GIBSON]

The theory sounded very promising in the beginning, but as said in the article, new software will start to support dual and quadcore cpu's to there would be no longer need for a technology like reverse hyperthreading

[vfrex]

That might be one of the most poorly written "articles" that I have ever had the displeasure of reading. I mean what, you just need a keyboard and an internet connection to get something published?

And Kougar, I think the idea is that if you can get all 4 cores working together with RHT, it would be pretty badass. Maybe I understood this mythical technology wrong though.

[GIBSON]

That might be one of the most poorly written "articles" that I have ever had the displeasure of reading. I mean what, you just need a keyboard and an internet connection to get something published?

And Kougar, I think the idea is that if you can get all 4 cores working together with RHT, it would be pretty badass. Maybe I understood this mythical technology wrong though.
It would be pretty badass TODAY, however in a year or so there should be a lot more software, OS's etc. supporting dualcore, quadcore and this would become an unnecessary technology

[vfrex]

Right, but it would be the perfect stopgap in the meantime.

[GIBSON]

Right, but it would be the perfect stopgap in the meantime.
Yes, it would. However do you consider it worth the amount of money that would be needed to do all the research. (if it is even possible?) And if so, wouldn't that money be better spent on the K8L?

[vfrex]

Maybe, maybe not. Do you use older software? Ever play older games?

I saw a funny post on slashdot that went something like "Writing multi-threaded code is like shooting yourself in both feet at the same time". The code is by no means easy to write. What if future AMD processors carried a compatibility mode which made use of the multiple cores, even in older software that doesn't support more than one?

Sure, cutting edge software development will support multiple cores/processors. But not all developers are going to be willing to add that level of complexity to the code, especially if speed gains from multi-core architecture could be had by throwing some optimizations to the compiler.

[Kougar]

That might be one of the most poorly written "articles" that I have ever had the displeasure of reading. I mean what, you just need a keyboard and an internet connection to get something published?

And Kougar, I think the idea is that if you can get all 4 cores working together with RHT, it would be pretty badass. Maybe I understood this mythical technology wrong though.

As far as the quality of the article goes, that is par for The Inquirer unfortunately. :???:

The only application of RHT that would interest me is with quad-cores, where I could turn it into a super-dualcore chip. For me, the entire idea with dual-cores is to be able to split the load so the system doesn't "lag out" running a single program, with a large boost in multithreaded application performance being a secondary bonus.

The bigger question is how much extra performance would RHT really give you anyway? It certainly wouldn't double the performance of a single threaded app, I'd want to guess around 20% on a good day?

[ToXic_WaSTe]

Maybe, maybe not. Do you use older software? Ever play older games?

I saw a funny post on slashdot that went something like "Writing multi-threaded code is like shooting yourself in both feet at the same time". The code is by no means easy to write. What if future AMD processors carried a compatibility mode which made use of the multiple cores, even in older software that doesn't support more than one?

Sure, cutting edge software development will support multiple cores/processors. But not all developers are going to be willing to add that level of complexity to the code, especially if speed gains from multi-core architecture could be had by throwing some optimizations to the compiler.

Well one of the 2 cores on a new cpu operates faster then a older one core chip from some time ago, so would still think it is rubish.