Rumor: AMD to launch super-fast 5 GHz 'Centurion' FX processor?

It used to be that a PC hardware enthusiast could buy the most powerful processor for his or her rig just by getting the one with the highest clock speed. That's no longer the case as new PC chips are more about additional cores and threads. This weekend, a new rumor claims that AMD might decide to enter the clock speed wars again.

Hexus.net reports via unnamed sources that AMD is planning to launch a new member of its FX processor line that has the code name "Centurion."  The article claims that it is based on the Vishera architecture and that only a limited number of these new Centurion chips will be made. While AMD's FX-8350 PC processor can run up to 4.2 GHz, the Centurion is supposed to run with a clock speed of 5 GHz. That's not only well beyond AMD's current fastest chip, it's also much faster than Intel's most powerful Core i7 processor, which only goes up to 3.9 GHz.

The article also claims that the price of this super-fast AMD chip will be $795. While that's on the high end, it's also pretty reasonable for a chip that has such a fast clock speed that's also supposed to be limited in number.

While clock speed isn't everything in a PC chip these days, there is something to be said about having the speed king bragging rights to a processor. It's possible AMD could launch the Centurion as a way to spoil the launch of Intel's Haswell lineup of Core processors in June.

Source: Hexus.net | Image via AMD

Report a problem with article
Previous Story

Why Nokia wants Instagram: Celebrities

Next Story

Microsoft continues Hyper-V growth, catching up to VMware

41 Comments

Commenting is disabled on this article.

Even though we have multiple core Processors the clock speed still has meaning , if it did not then all of us would still be using the 2 gig Processors and the makers would have just added more cores over the last years .
Now does the speed mean the same as it did years ago ? No , quite a lot has changed for sure . But the added clock speed with the multiple core processors will give us better processing and as I see it the faster they can go and the more useable cores the chip makers can give us the better .

I know that I am looking forward to see how this works , and mostly because I like to test and use the new products from both AMD and Intel .
Anytime we make a step forward it just makes technology move faster and new items are found that enhance our computers !

I certainly would buy it just for the lols of it.... but of course, my money won't give me the lols so... I will havbe to wait (thank godness javik isn't here)

Arceles said,
I certainly would buy it just for the lols of it.... but of course, my money won't give me the lols so... I will havbe to wait (thank godness javik isn't here)

But AMD FX processors run at 9000 more watts than Intel's!!

/s

While some have hard time accepting it we all need AMD to stick around even if it means this kind of "Limited" PR stunts. That's including Intel who needs AMD to stick around.

Hell I've been a red all my life but even I wouldn't buy this chip but it is grabbing head lines in the interwebs so it's basically a free form of advertising.

AMD has always copied something decent, overclocked it, and put on a better heatsink/fan combo because overclocking generates so much more heat.
maybe its a grudge but I'll never get over the AMD K6-2 fiasco. In my opinion the Athlon was the best thing they had going for them, but never again.

About 5-6 years ago I built my first quad-core computer with a Phenom. There were strange issues with it so I replaced every single part, but it had the same problems. Finally I replaced the CPU and it was fine. That was the last time I ever bought an AMD cpu.

More like he overpaid when he could have paid the right price in the first place.
AMD has run its course unfortunately they need to work on improving their architecture if they wish to continue to compete

Raw clock speeds are no longer what consumers need, I think it's pretty clear that clock speeds have been adequate for some time, the real issue these days is power consumption, heat, chassis size, and optimizations for multitasking scenarios (e.g. multiple cores, more threads, better multitasking with same clock speed).

Raw clock speeds do still have certain applications, especially with things like 3D rendering, but even 3D software these days are capable of exploiting the power of multi-core processors...

Multi core isn't just the future, it's the new normal. Phones have moved to dual core for power efficiency, and even the new Atom processors are dual core. The emphasis on clock speeds is irrelevant now.

AMD can throw out these numbers, but they don't as impress like they used to because people know that there is little value in having crazy high clock speeds.

The benchmarks will speak for itself.

I remember when AMD introduced their Athlon 500, I was a fan, I haven't used them since the intel "i" series. Multicore was one of the most noticeable differences in computing I'd ever seen. Now that I work with video, well everything matters. I guess if they can't quite win the benchmark wars, take what you can, in this case the speed "title". In the end, we need AMD to succeed to keep competition alive, from which we all benefit. I'm more interested in their new chips for mobile, maybe soon we can get a slim tablet that can play a few games too. (games not stupid birds)

Remember when the AMD fans derided intel for focusing on high clock speed over IPC? I'm guessing they have nothing to say now.

J_R_G said,
Remember when the AMD fans derided intel for focusing on high clock speed over IPC? I'm guessing they have nothing to say now.

You'd have a point if this was actually intended to be a mainstream product, in place of a limited run PR stunt.

Dr_Asik said,
They were right.

Didn't opine on that, just that I'm sure it'll be a different song this time 'round.

Awesome! I loved the days when you boasted about high clock speed to friends Would love to see something with only 2 or 4 cores with an even higher clock speed as standard though would be perfect for me.

ChrisJ1968 said,
but kick those speeds over 6 or more cores... insanity

Even the current FX's easily run at those speeds.
Even the FX4100 runs up to 4.7ghz with air cooling.

Yeah, i can't see them beating Intel in the foreseeable future as Intel has been faster than AMD since about mid-2006. because clock-for-clock Intel i3/i5/i7 line seems to beat pretty much anything AMD the last i checked (and generally uses less power to do it to) as unless AMD really boost the CPU's i don't think a Ghz increase will help that all that much as they probably have to be significantly faster than Intel in the Ghz dept just to get the same overall performance, putting cores aside. (hell, even with more cores that only helps in some cases)

because my i3-2120 (Sandy Bridge) was basically faster than all or nearly all of AMD's CPU's the last i checked in most cases and it's only dual core. sure, i suppose if you get something that really takes advantage of the extra cores (like say 4 or more) then AMD will probably beat my i3-2120 but lacking that clock-for-clock the Intel i got is most likely quicker. plus these Intel CPU's seems to be able to keep higher speeds without drinking a ton of power either as i am sure you could get some AMD CPU's in the ball park of Intel speeds (at stock Intel speeds) but probably require a decent overclock etc which just jacks up power consumption so in the end it's still all around worse than a stock clocked Sandy Bridge and the like.

p.s. in the end the best thing about AMD is that it keeps Intel from spiking there CPU prices too high. because when i got my i3-2120 it was just to appealing @ $115-125 price point (back when i got it in May 2012) as it was pretty hard to beat in that price range and was basically the best all around choice. but if AMD had not kept some pressure on Intel odds are those would be priced noticeably higher.

Edited by ThaCrip, Apr 15 2013, 5:29am :

LOL hardly. Maybe for people like you who probably do nothing but post BS on news sites all day. But some people do work and a quad core at 3.6GHz will be a lot faster than a quad core at 2.6GHz.

I have a hex core i7 at 4.9GHz and it still takes 2+ hours to do some of my 3D rendering.

ChrisJ1968 said,
that's good and all but speed is irrelevant now with the proliferation of multicores.

Its not totally irrelevant but its not linear either. The clock speed does matter; 4 cores at 5.0GHz are faster than 4 cores at 3.2GHz.

if game developers would quit being lazy and develop games for multi-core processors. but back in the day when we had only single cores, speed was essential. but if you have 4 or more cores running various threads for games, could be. I'm simply making a statement from a personal opinion.

Riva said,

Its not totally irrelevant but its not linear either. The clock speed does matter; 4 cores at 5.0GHz are faster than 4 cores at 3.2GHz.
It only matters when you're comparing CPU's from the same manufacturer and generation: ex. Core i5 2400 vs. Core i5 2500 (Both Core i5 Sandy Bridge CPU's). You can't cross compare clock speeds for something like AMD FX 8350 vs. Core i7 3770, because it's different architecture and it's apples vs. oranges.

ChrisJ1968 said,
if game developers would quit being lazy and develop games for multi-core processors.
It's not a matter of being lazy or not. Programming for multicore processors is very hard, and people are still just figuring it out.

There are tasks that parallelize well and others that don't. If the task is "painting a room", then putting 8 people to do it should be 8 times faster, because everyone can work with his own bucket of paint and his own wall. If the task is "cook dinner", then while 2 people can work faster than 1, it's not really twice as fast anymore, because there's still only one stove and it doesn't cook any faster than it used to. And if you try to cram 8 people in that kitchen they'll probably just step on each other's toes more than help speed us the process.

Games are very hard to parallelize efficiently, i.e. they're a lot more like cooking dinner than painting walls. Each game state iteration is a function of the previous, so you can't simply have each thread compute an iteration; you have to somehow subdivide the process into independent tasks that all access the same resources, some of which aren't any more parallel than they used to be (all I/O for instance). Proving a multi-threaded program is correct and debugging it is an extremely difficult task in general because the amount of possible "code paths" (all threads taken in consideration) grows exponentially with the number of threads, and on each execution the order of execution is chosen at random by the operating system, so you never really know what it's going to do.

I don't know where this idea of "lazy" programmers comes from, but I suggest you visit a game studio once in your life and ask how people work there. You'll learn that these people are usually brillant, highly motivated and that they work plenty of unpaid overtime.

Edited by Andre S., Apr 15 2013, 12:29am :

W32.Backdoor.KillAV.E said,
LOL hardly. Maybe for people like you who probably do nothing but post BS on news sites all day. But some people do work and a quad core at 3.6GHz will be a lot faster than a quad core at 2.6GHz.

I have a hex core i7 at 4.9GHz and it still takes 2+ hours to do some of my 3D rendering.

You do know 3D rendering highly benefits from additional cores right?

Dr_Asik said,
It's not a matter of being lazy or not. Programming for multicore processors is very hard, and people are still just figuring it out.

There are tasks that parallelize well and others that don't. If the task is "painting a room", then putting 8 people to do it should be 8 times faster, because everyone can work with his own bucket of paint and his own wall. If the task is "cook dinner", then while 2 people can work faster than 1, it's not really twice as fast anymore, because there's still only one stove and it doesn't cook any faster than it used to. And if you try to cram 8 people in that kitchen they'll probably just step on each other's toes more than help speed us the process.

Games are very hard to parallelize efficiently, i.e. they're a lot more like cooking dinner than painting walls. Each game state iteration is a function of the previous, so you can't simply have each thread compute an iteration; you have to somehow subdivide the process into independent tasks that all access the same resources, some of which aren't any more parallel than they used to be (all I/O for instance). Proving a multi-threaded program is correct and debugging it is an extremely difficult task in general because the amount of possible "code paths" (all threads taken in consideration) grows exponentially with the number of threads, and on each execution the order of execution is chosen at random by the operating system, so you never really know what it's going to do.

I don't know where this idea of "lazy" programmers comes from, but I suggest you visit a game studio once in your life and ask how people work there. You'll learn that these people are usually brillant, highly motivated and that they work plenty of unpaid overtime.


Nice and all, and might be true in some cases. But currently allot of games don't take advantage of threading while they easily could. And more often then not a PC game is a lazy Xbox port.
Supreme Commander is a game with beautifull support for multithreading, it does it wonderfully and you see the game sparkle and shine on multicore systems.

And yes its a game that highly benefits of threading due to the game genre.
But even FPS games. Quake engine and Unreal engine take lovely advantage of the cores. Even the old Q3 engine knows how to take advantage of multiple cores (not as good, but it does) and that engine was released many years before people even had multiple cores.
Battlefield 3 does horrible multithreading for example, a game not even a year old.
Neither does Dishonored or Tropico 4 and many others. While they easily could.

Shadowzz said,
Nice and all, and might be true in some cases. But currently allot of games don't take advantage of threading while they easily could. And more often then not a PC game is a lazy Xbox port.
How do you gather that they easily could? Have you read the source code of said games?

Shadowzz said,
Battlefield 3 does horrible multithreading for example, a game not even a year old.
Battlefield 3 is based on Frostbite 2 which is a highly scalable and multithreaded engine. It performs brillantly on all platforms it is designed to run on.

Dr_Asik said,
How do you gather that they easily could? Have you read the source code of said games?

Because there is no game that is just doing 1 task at a time unless it's perhaps pacman or tetris. Or similar very simplistic games.
But I would like an example of a modern game that benefits from shoving everything down the throat of one core over multithreading it.
Dr_Asik said,
Battlefield 3 is based on Frostbite 2 which is a highly scalable and multithreaded engine. It performs brillantly on all platforms it is designed to run on.

More then often it seemed slower then it should be though. I am unknown with frostbite details. But that's how it felt while playing.

Athernar said,

You do know 3D rendering highly benefits from additional cores right?

No i had no idea, I bought a hex core CPU because i like the number 6. /s

And you do know that the speed of each core also matters right? The whole reason i've massively overclocked it.

Shadowzz said,
Because there is no game that is just doing 1 task at a time unless it's perhaps pacman or tetris. Or similar very simplistic games.
That doesn't mean that these tasks are easily parallelizable, especially when you need to gather all results of all computations in one coherent state 30 times per second.

But I would like an example of a modern game that benefits from shoving everything down the throat of one core over multithreading it.
What's the point? Obviously spreading calculations over multiple threads, if done intelligently, will improve performance. There's no performance benefits to not doing that, but that doesn't make obvious how to take advantage of multiple threads in a game.

Riva said,

Its not totally irrelevant but its not linear either. The clock speed does matter; 4 cores at 5.0GHz are faster than 4 cores at 3.2GHz.

clock speed is irrelevant. its all architecture. clock speed only tells you how fast a flip flop latches in the chip. a cpu could take 4 clock cycles for an instruction, while another chip could do 2 instructions in one clock cycle. both of these chips running at the same clock speed will have different performance. the second one will be 8x faster.

W32.Backdoor.KillAV.E said,

No i had no idea, I bought a hex core CPU because i like the number 6. /s

And you do know that the speed of each core also matters right? The whole reason i've massively overclocked it.

While true that speed somewhat matters it's not the biggest factor by any means.

You could suicide overclock current FX's and they would still get their ass handed to them in single-threaded workloads on stock Intels. (More or less I guess )

One of AMD's biggest problems is automated arch design, it'll NEVER be as good as someone actually going trough bits. They need to drop it, of course it's cheaper and all but in the long run they will loose money with this crap.

Riva said,

Its not totally irrelevant but its not linear either. The clock speed does matter; 4 cores at 5.0GHz are faster than 4 cores at 3.2GHz.

I hope you're trying to make a joke with this. The 5.0 Quad system could end up performing MUCH slower than the 3.2 Quad system. GHz / clockcycle doesn't translate to real performance.