AMD says its badly needed quad-core desktop processors are on the way, and they’ll arrive bearing a new name. Two quad-core chips will be available in the second half of the year, the Phenom FX and the Phenom X4, and a dual-core chip based on a similar design called the Phenom X2 will also appear by the end of the year.
AMD Barcelona K10/K8L chip is known for very long time.
It is a great designed CPU but AMD should release it much earlier then that.
Beyond bragging rights, quad core means 0 to 99.9% of all desktop users out there.
The enthusiast market, as noisy as it is, accounts for a minute part of the market share.
The number of applications that can scale nicely on two cores is slowly growing… but literally zero desktop applications come even close to loading 4… or 8 cores.
AMD needs a quad core Opteron CPU for workstations and servers -NOW-.
Not next month, not even tomorrow.
NOW.
Hopefully AMD will release the QC Opteron parts on time (End of Q2/07) and -in volume-.
– Gilboa “typing this on a dual – dual Opteron workstation sitting next to dual – quad core Clovertown workstation” Davara.
I’ve posted about this before, but not lately. So here goes.
My largest XDMCP desktop server services about 50 users. We benefit from its 2 3.2GHz Xeons a good bit of the time. But not all the time. And that’s with 50 users. The average desktop has dozens of processes running, *ALL* of which are asleep at most times.
By that I mean that most users are not using any processor at all at any given time.
With 50 users, we *might* benefit noticeably from a quad core.
But from a marketing perspective, I agree that AMD needs to get a quad core out fast.
Sad, isn’t it?
Edit and P.S.: To those about to mention specialized workstation uses, I agree completely. There do exist individual apps which can benefit greatly from multiple cores. It’s just that most peoples’ desktops don’t
Edited 2007-05-15 17:51
Yeah, but the Opteron, as the Xeon target both workstation (E.g. 3D modeling, heavy math, etc) and servers. (Server applications are designed to scale very nicely)
Having said all that, even server applications (as my own – at least thus far) tend to start being memory/bus bandwidth capped on Clovertown/2x4C machines – Making it less effective then say, a 4x2C Opteron 8xx machine. (With 4 independent DIMM banks)
In short, 4C is nice, but hopefully AMD’s Barcelona on-die memory controller will yield better results then Intel Clovertown’s northbridge.
(…Intel’s CSI is not far behind)
– Gilboa
I’m currently banging my head writing out my doctoral thesis and all I can only say that more processor power is welcome. I’m not the most organized person, and as I’m writing I see the kind of results I need, or results of stuff that I did in my early years is incomplete/inapplicable and I need the most up-to-date results.
So what does this mean? On a typical day, I’ve got Matlab running simulations in the background, LaTeX and the assorted TeX stuff running up front, some diagramming tool, a web browser, code editor, and maybe some music
This is handled well by my Core Duo processor. One core is pretty busy with Matlab, the other cores are more than capable of doing the rest of the tasks. Why am I keen to see more cores? It means that I can run more instances of Matlab at once! Most of my stuff is done in batches, so assigning each core its own task works. Bring on quad and oct cores!
Yes, I know I could split the workload between multiple machines, but it’s much easier to ask for a machine upgrade than asking for 4 new machines.
If Barcelona isn’t ready then it isn’t ready. Unlike Intel, AMD is unable to get away with relasing unfinished hardware.
Any numbers?
It seems that AMD’s research department is only concerned with beating Intel at its own game. This is foolish, IMO. AMD is doomed to always be a follower unless its engineers can come up with a revolutionary new CPU architecture based on a revolutionary software model. The new architecture must address the two biggest problems in the computer industry today: reliability and productivity. Unreliability puts an upper limit to how complex our software systems can be. As an example, we could conceivably be riding in self-driving vehicles right now but safety and reliability concerns will not allow it. Why? Because there is something fundamentally wrong with software. Fortunately, a software model that solves these problems already exists. It is called the “non-algorithmic, synchronous, reactive software model. That’s what Project COSA is about.
Project COSA:
http://www.rebelscience.org/Cosas/COSA.htm
> AMD is doomed to always be a follower unless its
> engineers can come up with a revolutionary new CPU
> architecture based on a revolutionary software model.
> […]
> a software model that solves these problems already
> exists. It is called the “non-algorithmic, synchronous,
> reactive software model. That’s what Project COSA is
> about.
Languages/systems like COSA need much more research before well-performing CPUs can be made for them. Most of the current research is focused on embedded systems with completely different goals than general purpose languages on standard processors.
“””
Languages/systems like COSA need much more research before well-performing CPUs can be made for them.
“””
While I am not familiar with COSA, this thread brings to mind something that I have noticed about this industry. Sad to say, but having a great product that is years ahead of its time is usually a death sentence. Although it also virtually guarantees you a permanent place in the history books, and an ongoing presence at sites like OSAlert.
In extreme cases, you can even have your grand old name dragged through the mud for decades, like Amiga.
Edited 2007-05-15 15:55
Languages/systems like COSA need much more research before well-performing CPUs can be made for them. Most of the current research is focused on embedded systems with completely different goals than general purpose languages on standard processors.
I agree. All it will take, however, is a working COSA OS and development tools that a lot of people can play with. That will come. The theory is sound, IMO. COSA would be perfect for embedded apps, especially mission-critical systems such as is common in the automotive and avionics industries. In this light, I think that AMD would do well to branch heavily into the embedded cpu business. Coming out with something that leaves everyone else in the dust would certainly help. That’s where COSA comes in. But then again, I’m biased.
“AMD is doomed to always be a follower”
Actually, AMD has been leading the CPU landscape for quite a while now:
1. They created x86_64, and Intel was forced to use that technology in it’s 64bit x86 products, called EM64T.
2. They had the first successful dual core x86 processor, Intel only came out with one after AMD did, and at first they were 2 separate dies in the same package. Core Duo came later.
3. They have been using Hypertransport for a while now, and Intel has not caught up with that technology, they are still stuck with the FSB, while they try to get CSI working.
4. Hypertransport allowed them to develop Fusion, which will allow gpus to be plugged into a processor socket on the MB, which will then run on the HT bus. Fusion also lets AMD build multi core chips where one or more x86 cores are replaced with gpus, physics processors, you name it. Intel just released details of thier similar plans, which are not as far along.
5. For the longest time, AMD was the only one of the 2 with an integrated memory controller, which allowed the Athlons to keep ahead of the P4 in terms of speed, as it didn’t have to talk to the MC on the northbridge. Only with the Core Duo was Intel able to negate this advantage.
Edited 2007-05-15 15:44
Actually, AMD has been leading the CPU landscape for quite a while now:
I agree. I did not mean a follower in the technological sense but in the business sense. Intel can outspend them anytime.
I agree, Intel can outspend them anytime they want, but that is mostly because of thier huge size, and diversified product portfolio. AMD has taken steps to remedy that by buying ATI, which gives them roughly the same capabilities as Intel, so hopefully they can compete on more equal footing.
Isn’t both core duo and core 2 duo more or less separately kernels still? They still share data via cache and fsb, not core to core, or?
(e.g., IF/THEN/ELSE, WHILE, GOSUB, GOTO, DO WHILE, DO UNTIL, etc…)
Please not back to begin eighties…
That article clearly has some good points but also displays some lack of real insight by ignorant and false statement such as “…hardware is inherently synchronous…” and similar statements regarding the brain.
Anandtech has some interesting information about AMD’s future plans in their article:
“AMD – The Road Ahead” by Anand Lal Shimpi
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2986
They seem to think that also AMD – especially after the Ati acquisition – may still have many new innovative and competitive CPUs and technology in development. Time will tell?
I hope the new chips are not DRM infected.
http://www.infoworld.com/article/07/03/28/14OPcurve_1.html