Dual Core Intel Processors Sooner Than Expected 257
Hack Jandy writes "AnandTech reports that Intel's Smithfield processors are going to get here sooner than they originally predicted; most likely within the next few months. Apparently, the Intel roadmaps reveal that the launch dates for next generation desktop chipsets, 2MB L2 Prescotts and Dual Core Smithfield processors (operating at 3.2GHz per core) are almost upon us - way ahead of the original Q4'05 roadmap estimates. Hopefully, that means Intel will actually start shipping the new technology instead of waiting four months after the announcement for retail products."
Programs (Score:5, Insightful)
How is this year going to be different?
Even if you *could* get SMP aware versions of your software, would it be worth it? Lots of problems are harder to solve when you add SMP to the mix.
Gamers will be put off by the fact that games can't take advantage of SMP.
Home users will be put off by the fact that their $500 Dell surfs the world-wide e-mail just fine.
Buisness user may take advantage of this in servers, but there's only so much cooling and power you can provide to a 1-U server.
So, how is dual core going to ever be anything bigger than Itanium, Xeon, or any of the other technologies that fail to meet customer expectations?
The 80s called about "multi-tasking". (Score:1, Insightful)
You don't generally run one application at a time, right? So I don't see the problem.
Don't print useless press releases! (Score:3, Insightful)
Want to change Intel's behaviour? Don't give them any press when they announce "real soon now" stuff, only when they actually ship. But if /. (and other media) print every press release, the press releases will keep coming.
my epiphany... (Score:5, Insightful)
Both Intel and AMD have decided upon dual-core as the future of desktop computing. There will be no more massive Mhz increases... instead the focus is now on parallel computing.... But, seriously, how many CPU intensive applications outside of the server arena take advantage of SMP?
As someone who has ran dual-cpu workstations for years, I can personally attest to the fact that 99% of CPU heavy tasks do not make use of SMP.
Think about it... That copy of Doom3 or Half-Life 2 that you just bought, that runs like shit on even top-of-the-line hardware, isn't going to run any better on Dual-Core, because these games are not designed to run multiple threads simultaneously. Neither do most archival programs (WinAce, WinRar, WinZip, SevenZip, etc etc). Nor do many of your encoding tools (though FlaskMPEG and GoGo-No-Coda are noteworthy exceptions).
As a geek, I can attest that the *nix arena isn't much better. Just because the source is open and available does NOT mean that the author(s) ever considered coding CPU intensive tasks for multiple processors. And "porting" tasks from single threaded to multiple threads is NOT a simple task. This is one of the reasons that there are Computer Science degrees -- writing good SMP code isn't something you learn at technical schools (or even half the full Universities out there).
Don't get me wrong... as someone who has ran SMP boxes for the past 10 years, I'm really excited about Dual-Core. But don't expect it to be worth a whole lot for the immediate future... as no one outside the server arena really codes for SMP.
Re:Office use? (Score:3, Insightful)
Sure, the speed-up isn't nearly as large, but having a spare core sure would prevent many slowdowns.
Nice for some apps. (Score:3, Insightful)
Its a bad move IMO on AMDs and Intels part - personally rather than head to dual cores I'll be looking more and more towards how to get the maximum (i.e. overclock) out of the higher rated single core processors - and this is from someone who normally upgrades every 12-18 months.
That said if the dual-cores overclock well my stance may change....
Re:my epiphany... (Score:3, Insightful)
Re:Just to be clear... (Score:4, Insightful)
I don't know what's up with Intel lately. They're giving too much away in the x86 market to AMD, and they can make good processors (P-M, for example).
Re:Bleh... (Score:3, Insightful)
Re:Office use? (Score:3, Insightful)
Re:Programs (Score:3, Insightful)
No - you said it yourself, an area where extra CPU power is useful is video work. More and more people are getting digital camcorders, and want to transfer the movies to PC to email to friends, burn to DVD, or whatever. Ordinary people are going to ask for PCs that are "good at video".
Further, if the CPU manufacturers move exclusively to dual-core procs, where are the OEMs going to get single-core ones from? Make no mistake, if Intel and AMD want this to happen, it will happen, and everyone else will either have to go along for the ride, or build their own processors.
Re:Programs (Score:1, Insightful)
While you might or might not know much about it, its a great platform to write software for. A lot of future software will be written for it, that will most likely include most Microsoft software.
You get twice the cache too, on Opterons you get close to twice the subsystem bandwidth! (Not when accessing the same resource / accessing non shared resources through the other processor (ht link))
Office systems dont need this kind of thing. This systems have a low average performance usage with rare spikes.
This is where I think architectures like The Cell come into play. Processors could theoretically share some workload with other machines.
Most office tasks requiring some performance, like excel-calcs and db searched could easily be paralized. Think of all the thin client nodes giving all their unused cycles to the backend.
A beowulf... aww crap, need my medication.
Re:But will they be 64-bit? (Score:2, Insightful)
Re:my epiphany... (Score:5, Insightful)
What we have here is simply the fact that, as always, software is years behind the hardware it runs on. This is a classic chicken-and-the-egg situation. "There's no SMP software, so why by a dual?" vs. "Nobody has SMP hardware, so why write SMP-aware apps?".
Thankfully, there are many SMP-aware apps available, not even getting to the fact that with single-threaded apps on SMP you can for example encode video and do other CPU-intensive tasks simultaneously and at their "native" speeds.
Games are probably the worst example to use for touting SMP benefits because they are written with the single-CPU mindset. This is a software shortcoming, yet many posters see this is a flaw of SMP? Silly. If you're using games as an SMP detraction, then you're not the target for SMP until the software is written to take advantage of SMP. Again, this is a software shortcoming, not a hardware flaw.
Then we have the "well office-type users have no need for SMP". Well, that may be true, but so is the fact that office use does not require >1GHz CPU's, yet offices are filled with >1GHz machines. The nature of the "CPU business" is such that your products must constantly improve, or you will soon become irrelevant. You can only make CPU's run so fast in the physical world, so after you've wrung all the easy MHz gains out of a process, what's the next "easy" gain? Parallelism. We don't expect Intel, AMD, et al to just say "Well, that's it, we can make them no faster", do we? Heck no. Instead of more MHz, we now have more cores. The software will follow, and in the meantime the hardware is usuable now.
The fact of the matter is this: there are real, physical limitations to the manufacture of ever higher speed CPU's. We're going to hit the brick wall shortly using current processes, so the next logical step is to parallelize the CPU. If you can't make 'em faster, then you divide and conquer.
As someone who runs a few SMP systems, I, for one, welcome our dual-core overlords. So I can run dual-core? Heck no, that's for the gamers and office-workers
This will lower the barrier of entry for SMP use for the masses. After they are dragged, kicking and screaming to SMP, people will notice a smoother, more productive computing environment. Also, us dual-CPU folk can now move up to quad cores with relatively little additional expense. As SMP moves into the mainstream, the software will follow. Any programmer worth his salt knows that it is trivial to parallelize many compute intensive tasks such as media encoding/manipulation, imaging, rendering etc. Now that the hardware is (almost) here, the apps will follow.
I am sincerely interested in hearing any response to these points I've made.
Re:Nice for some apps. (Score:5, Insightful)
Tough. Chip makers are up against a technology barrier right now, and clock speed increases in the CPU don't make RAM or disk or interconnect faster anyway. How about just putting a 4MB cache on-die? That wouldn't require a massive clock speed increase but it would speed things up. I'm not an EE but I'm just pointing out that there are many, many things that have been left in the dust by Moore's law that could catch up and make quite a difference. Does your computer have 4+GB of DDR memory? ATA-133 drives with 8MB cache? PCI-X? A 64-bit CPU and an OS that knows how to use it fully? In what other ways are CPUs waiting on everything else, that could be improved to make things run faster overall?
Learn to parallelize your code where possible. Optimize your existing code. Software optimizations yield stunning improvements compared to incremental clock speed bumps anyway, and (unlike hardware) affect every installation of your app.
>Its a bad move IMO on AMDs and Intels part
OK genius, what's the alternative? No improvements in processors for years, until somebody makes a breakthrough that enables 4+ GHz processors? What happens when they hit the next roadblock?
Hardware has been so far ahead of software for so long that we've become accustomed to solving bloat with "just buy a new computer". It wouldn't kill us to spend a little time profiling code. The economics have been (in many cases) such that it just made more sense to throw money at new hardware. If that no longer makes sense, throw money at software optimizations for a little while. It doesn't exclusively mean that we have to force every algorithm to operate in parallel. It could be as simple as releasing fat binaries of apps that are compiled to target recent CPUs (no more shipping 386-optimized code to every customer), or *gasp* writing more efficient code in the first place.
Re:my epiphany... (Score:4, Insightful)
Somebody mod this guy down, he's talking out of his ass, and does not deserve an "Insightful" mod.
Sorry if that sounds harsh, but he really doesn't know what he's talking about. He should try running a dual-cpu box before he makes comments on the state of software and SMP.
Re:yada yada yada (Score:2, Insightful)
On of the main reasons that P4s have deep pipelines is to compensate for the fact that memory speeds have not generally risen anywhere near as much as the chips ability to consume the memory bandwidth. This problem doesn't go away with your multiple 3.8Ghz 386's - if anything the problem is compounded (much as it will be when there are two P4 cores sharing the same memory bandwidth as a single P4) so your 386 will sit there spinning it's little 386 wheels everytime it needs some data or cocks up a branch prediction.