Slashdot Mirror
Intel's P4 3GHz w/ 800MHz Bus & Canterwood Chips
OldGrayDave writes "Intel steps out today with their new Pentium 4 3GHz chip that runs on an
800MHz System Bus. They've also released "Canterwood", the chipset chipset
for the P4 that supports Dual Channel DDR400 memory, native Serial ATA 150, RAID 0,
AGP8X, USB2.0 and a host of other bells and whistles.
Check out this showcase and performance analysis at HotHardware, to see what
all the buzz is about. Intel distances themselves again from the Athlon." Or, you can read more at Hardavenue, mbreview, Tom's Hardware, hardware unlimited, or The Tech Report. I dunno...hardware gets faster, bus gets faster. Tide goes in, tide goes out.
I have to wonder what the point is with some of these new faster processors. At this point, almost no applications can really take advantage of the fastest chips available. My sister uses a 500 MHZ machine at home, and as far as I can tell she has no real issues with its speed. I have to wonder if Intel is just shooting itself in the leg, spending needlessly large amounts on R&D to produce chips that no one actually needs. PS - FP?
This isn't meant as a troll, but I'm sure many of the more... sensitive ... Mac users will take it as such anyway.
This speed boost is great for the x86 world. Speed keeps getting better. Intel and AMD keep competing and leapfrogging each other to greater heights. My sorrow is that Apple's offerings really are *years behind* right now. I know, I know, speed doesn't matter when Macs are slower, but when Macs had the speed advantage, the Mac users claimed speed was all-important and there was no problem attacking the PC users based on their sorry speed. Mac users, like everyone else in the world it seems, aren't objective - if the PowerBook is thinner, they claim size (ahem) is important. When the PC world shows us the Superthin Vaio, we say that Size doesn't matter, it's how you use it (ahem again). And that's the problem; that's why Apple doesn't feel the need to force speed increases out of Moto and IBM to keep up with the Joneses - Mac users are so damn faithful, that they don't apply any market pressure to Apple to force them to compete! Instead, the "Mac Faithful" DEFEND Apple's weaknesses, allowing Apple to slack off in the processor department.
Next time a MacZealot defends Apple's 1 Ghz processors on a slow bus, tell him that he's NOT helping Apple. The way to help Apple is to absolutely demand faster processors, and threaten to switch to x86 if they don't deliver. If we give Apple a "Get out of Jail Free card" with regards to processor speed, we'll NEVER be competitive with Intel.
And yes, I've heard the RUMORS about the IBM chips. They'll still be far behind this, RISC or not.
Someone else (Gothmolly) said 'I want cheap SMP, not more MHz.' earlier on. I thought about his reasoning for a while and have to suggest at least considering RAIC as a way to get cheap SMP. Redundant Array of Inexpensive Computers - similar in thought to the original purpose of RAID (RAI Drives) - use a few cheap pieces to get the same or better performance of one wickedly expensive piece of hardware.
... nice. Honestly though I am way more excited about the SATA/RAID 0 performance than I am the additional CPU horsepower. Sustained serial reads of 96MB/s and sustained serial reads of 86MB/s - not burst but sustained -:- DAMN. That is easily twice the performance of my current rig, possibly three times the performance. I care less about a CPU running 7% faster; I wouldn't even notice the extra 40fps on Quake3Arena going from 410fps to 450fps but the ability to move data back and forth to the hard drive three times as fast is going to make this one machine worth upgrading to.
Today, if you were to search around a little to scrounge up one of the 15% coupons floating around on the net (www.fatwallet.com for example) you could go to Dell and have a fully loaded system (Celeron 1.8GHz, 128M DDR266, 40G IDE, 8M onboard video, Intel Gigahertz NIC, 48xCD, keyboard and mouse) delivered to your house for $240 after rebate ($100 rebate but they are pretty good about paying them.) I think you can upgrade the hard drive to dual 80G drives (buy one get one free if you upgrade) for another $60, bringing the total price to $300. Add a $50 two port KVM (I use the Linksys, has build in cables) to your existing rig and now your monitor, keyboard and mouse can instantly switch between the two systems. Have a massive process that hogs the CPU, swap over to the other machine to do whatever you want while it runs. I have been doing this for a while and the ONLY drawback I have seen so far is not being able to cut and paste from one to the other. Other than that they are effectively one machine with two discrete workspaces.
As for the new hardware
IMHO the advances in hard drive performance are the real story here. Running the P4/3G on a 400FSB vs the old 333FSB is nothing compared to getting 3x the performance from the drive subsystem.
Glonoinha the MebiByte Slayer
Is there a limit as to how many controllers can be on board? Could someone for example put 12 controllers on one board with not much trouble?
This P.I.G. will walk on the water, This P.I.G. will walk on the sea, This P.I.G. will walk whereever he wants.
I've reclocked my Athlon 1700+ (TBred core) to 8x202MHz (404MHz DDR) on my ASUS nForce2 chipset board, using a single Corsair PC3200C2 DIMM (yes, two DIMMs would be better, but they were too expensive at the time). It's just a matter of selecting the right BIOS settings. I left the voltage levels at their defaults. MemTest86 verifies that the memory is stable at that speed. Red Hat Linux runs until I need to reboot for the usual kernel/glibc upgrades. I went this this approach because I wanted to optimize the performence/power consumption balance, what with the machine running 24x7 and all.
Of course, tweaking speeds like this is not guaranteed to work, yadda yadda, but it generally does if you built your system right.
If you want serious firepower, build a dual Athlon box, which should cost no more than the uniprocessor P4 being reviewed. time make reports a bit over 9 minutes when building Wine with MAKEFLAGS=-j2 on my dual 2400+ (not overclocked). Nice, especially when you forget the --with-nptl switch the first time around (d'oh!).
Of course, next week, the Opterons ship, starting with Opteron DP 240's and 242's. It's unclear whether there will be cheap workstation motherboards available right away or just the seriously nice (and expensive) Newisys-designed 1U rackmount servers. It appears that AMD is going to use the Opterons to slap the high-end P4's around, saving the Athlon 64 until they want a low-to-midrange 64-bit desktop platform. I'm surprised the various hardware site reviewers haven't picked up on this.
I think the "make -j4" example is a good example of using all your SMP processors for maximum performance, but not really why I prefer a dual system to a single processor. Frankly, one P4 3GHz is still gonna significantly outperform a dual 1.5GHz Xeon system even for this operation..
I have a dual Celeron 533 MHz at home, and while it's by no means a speed demon, I still prefer it to a single processor system of even 3GHz class.
The reason: No matter how processor intensive a background task I may be running, my computer continues to be smooth and usable. And if it's a long-running task, this is especially important. While it might be nice to be able to run the background job in an hour instead of six, if I cannot use my computer for that hour, I'm actually significantly more inconvenienced.
Yes, I could have two systems and a KVM, instead. But really, SMP is so much less cumbersome. And Intel's Hyperthreading does not provide this benefit, so the next system I'm hoping to get will be a dual Opteron.
Peace and love, y'all
All that stuff about huge speed increases and sackloads of extra memory bandwidth, reduced clock cycles, RAID...but when you eventually get to the performance testing it seems very little faster than top end current boards. Perhaps if you have a daily compute-intensive job that is slowly growing and currently takes 23hours, you would get excited, but as a developer I guess I might gain a few minutes off my build times (and that's staring into space thinking time anyway.)
I'm not knocking progress: the lower voltage and the ability to use a 4-layer board, plus the serial ATA on-board support look nice, but the number of people with more money than sense needed to get a fast R&D payback isn't that high at the moment.
Or is this a cunning plan to make money through selling compute farms to rogue states that have just decided they need WMDs really fast?
Panurge has posted for the last time. Thanks for the positive moderations.
"But my 8088 scrolls text just fine! Why would anyone want something more powerful?"
Because there is neat new shit that takes more power. Receant example: HD multimedia. The Windows Media 9 HD demos kill a P4 1.6 and really take at least a P4 2.4 to play well. Means that if you have a 2.4, which is pretty good these days, that is about ALL your system can do, not much room for bacground tasks.
Or how about speech recognition? There are some nifty new technologies in speech recognition the integrate it better with text parsing for far more accurate recoginition. One problem: they take loads more power than normal speech recog, which takes a bit itself. Given that ideally this should happen in the background as a normal part of the OS, more power become critical.
Or how about better game AI? I am so sick of 3d bots that get "good" by becomming aimbots or RTS AI that attacks you in teh same predictable way every time. I want smarter AI. Well, to do that it is going to take more processing power. Teh smarter the AI, the more CPU time it needs. all this while still doing all the other calcuilations a game needs (like physics and game logic).
We are not even close to expending the need for more computer power. As power grows, we'll simply find new and creative way to use it that were not before possable.
After all, my 8088 scrolled text like a champ, but I much prefer my P4.
The people who made VisiCalc had to worry about fitting the program on tiny computers; people making spreadsheets now don't, and we can use lots of graphics, features, higher level languages, maybe embed python/tcl/scheme/whatever, and generally have a lot of fun. The VisiCalc programmers actually thought about how much memory the character screen would take.
I see the usual replies claiming that 1GHz is more than fast enough and the usual critical replies citing video editing and speed recognition. Now while I can agree that more speed is often good, are these really examples where it makes a lot of sense to buy hot, expensive, and power hungry general purpose CPUs to handle special purpose tasks? 3D graphics didn't wait for processors to hit 20GHz, but a 300MHz graphics processor can outrun any general purpose CPU. Video editing is another good example. Why is it so slow? Because it involves compressing lots of data. A team of graduate students could create an FPGA that runs rings around a P4 for video compression (by, say, a factor of 20). Speech recognition is the same way.
In short, paying $1000+ for a processor that's 9% faster and uses 15% more power is not a good solution for "I need more power for video editing," especially when you should be able to get 20x-50x performance increases for 10% of the cost.
collection of linux boxens: check
add compiled kernel with mosix patch to grub: check
remote display pointing to system with monitor, kb & mouse: check
reboot.
You've got to be real careful with those Dell prices, as they are very misleading. The front page price is after rebates and minus OS license (despite the fact that they won't sell you a computer without a Microsoft (P)OS on it). Also, there are usually components and applications bundled in that you don't need or want, but can't choose to exlude from the order. Just go into the customize wizard, and you'll see the pricetag start to inflate to reality.
I don't know which system in particular you were looking at (because you didn't provide a link), but I just went to Dell, and the cheapest DP system in the cheapest config (2x 2GHz, 256MB mem, etc) is US$1397 (after rebates). Step up to a DP 3GHz, and you're looking at US$3097 (once again, after rebates). Now, today I was curious, so I went and priced out 2 3GHz Xeons, a high-end Tyan m/b, and 256MB of memory. It rang up to around US$2250 (cash, no rebates). So the Dell is not that well-priced when you consider that they are then charging $847 for a mid-tower case, a 20GB hard drive, a CD-ROM, a floppy, a keyboard, a cheap mouse, and a low-grade video card (about US$225 in components). And you aren't getting a motherboard anywhere near as nice as the Tyan. Just to be fair, I also priced out the same config at 2GHz (the min), and it turns out they don't screw you quite as bad, only charging you about $497 for the rest of the bundle, but that still isn't a bargain at over double the off-the-shelf price.
It gets really pricey when you consider that I already own a case, a floppy, a much nicer keyboard and mouse, a nice DVD/CR-RW, and a hell of a lot nicer video card. So I'm paying US$497-847 for inferior crap that I don't need. Also, given the fact that I am currently running a dual-proc 1GHz machine with 1GB of mem for my primary workstation, upgrading to anything less than the 3GHz solution with 1-2GB of mem doesn't make much sense. And to build a system comparable to what I have now, and upgraded with DP 3GHz + 2GB of mem, using all Dell parts (so I don't have warranty issues later when I call for service) will run about US$4950. That's about double what it would cost to upgrade my existing system and recycle 80% of the parts I already own.
So why again would I want to buy this? Oh, right. Because it's easy as Dell. Just like my Inspiron 7500 that cost me US$4500 and that's averaged a major service every 6 months since I bought it.
Sorry. I've been down that road before.