Slashdot Mirror
ASUS P4 Motherboard Bests Intel, Says Sharky
ravedaddy writes: "The Pentium 4 has been out for a few weeks now but with only Intel's own motherboard having come out upon release of the P4, the choice was limited. SE has reviewed of the first Intel 850 based motherboard from ASUS the P4T, which is actually faster than Intel's own. With features including the i850 chipset, a 400MHz FSB, four RIMM slots, ATA/100, five PCI slots and AGP Pro 4x, the ASUS P4T looks formidable. Using this new board, the authors were able to overclock the Pentium 4 1.5GHz easily up to 1.68GHz." Does it seem like the 2nd GHz mark is approaching a lot faster than the first one did?
First Linux, now Intel. The P4's showing *everyone* up.
While the original was really just a predicition, Intel's marketing and engineering plans depend on doubling the performance/price of the CPU every 18-24 months ot ensure a healthy stream of upgraders and new applications. (Substitute clock speed for performance as necessary.)
Call it Moore's Law of CPU Marketing.
--
Business. Numbers. Money. People. Computer World.
If only it were that easy. In most bioses the process of assigning specific IRQs to specific slots is extremely convoluted and way over the head of joe average.
The average user who has a friend or company set this up for him is completely screwed if his bios loses its memory for any reason. At least with jumpers/dip switches the settings need only be set when the hardware is installed and they stay put thereafter.
---
Where can the word be found, where can the word resound? Not here, there is not enough silence.
"Where shall the word be found, where will the word resound? Not here, there is not enough silence." -T.S. Eliot
It happens all the time. I think the latency's a bit lower going through continuity modules but it's still there.
---
Where can the word be found, where can the word resound? Not here, there is not enough silence.
"Where shall the word be found, where will the word resound? Not here, there is not enough silence." -T.S. Eliot
wince
Put the pun on the ground and raise your hands in the air! Back away from the pun! This is your final warning!
Maybe the state's highest function is to grind out insoluble problems. (Zelazny, Hall of Mirrors)
I own a Savage4 card. One of the main reasons I bought it is because S3 chipsets don't need an IRQ...
People replying to my sig annoy me. That's why I change it all the time.
The entire Intel line of processors is sadly inefficient... and capable of replacing the halogen burner on my stove. The performance of the P3 and P4 is NOT on track with Moore's law. The Megahertz are because they want people to think they're doing well... but when you reduce efficiency per clock cycle, but up the number of cycles, your taking 3 steps forward and 2 steps back. Nice try intel.
"Politicians find new names for institutions which under old names have become odious to the people."
The StrongARM chips also don't have FPUs... not that one is really needed on a cell phone, but hey, if you are going for that much power in a cell phone, you might as well be able to play text-mode Quake...
--
"It's tough to be bilingual when you get hit in the head."
It's not a trick. You don't notice processor speed in terms of difference, but in terms of a factor. So if you overclock a processor by 50%, that is noticeable. If you overclock it by 10%, it is not, even if the difference in clock speed is the same in both cases.
EagerEyes.org: Visualization and Visual Communication
Because computers most certainly haven't gained exponential speed increases over the last 10 years. If that were the case systems would be more than 32 times faster than ten years ago. What we have is a huge increase in marketed clockspeed, but I don't see a significant increase in system performance after Intel separated internal CPU clock speed from the system bus. Do you?
--Maynard
Read the article. Don't jump to conclusions. It's a 100Mhz FSB that's quad pumped (like AGP4x). They bumped the FSB to 105Mhz but the RDRAM couldn't handle it so they changed it to triple pumped to make it work.
Because he wasn't talking about nature, he was talking about technology. And that is astounding. Prior to computers, I can't think of a single technology that has exhibited exponential growth, can you? Perhaps destructive yeild of nuclear weapons, though I don't know enough about that to say for certain. Output from power plants? Maybe, but that's probably more a factor of increased demand than increased technology, they're just building 'em bigger. Anybody got counter examples?
"The question of whether a computer can think is no more interesting than that of whether a submarine can swim" -EWD
My mother is still using a 20Mhz mac to surf the internet... It's amazing. A 2Ghz chip is imminent...100 times the clockrate of my mother's lowly computer. And in another 18 months, it will be 200, 400, 800,...then the killer robots! Gotta love an exponential curve!
Ryan Finley
Ryan Finley
SurveyMonkey.com -- Create your own professional surveys
It seems to be their intent to distribute a cpu, chipset and a reference platform. Individual PC manufacturers either draw up their own boards or buy them from 3rd parties based on the reference platform. This has always been the case. I think there would be legal issues if Intel decided to own the entire PC design. It would effectively limit competition and open up the opportunity for lawsuits. Of course, IANAL.
And quit your bitching about "fitting" a 400MHz chip in your cell phone. In a couple years, you might see a 1GHz Xscale in your phone. What.. you gonna complain that it's too fast for your phone then? You dumbasses are never happy. Too slow. Too fast. Too hot. Too cold. Too Intel. Too AMD. Just shut the hell up and enjoy the ride already.
The article does make reference to an Athlon on a 760 board with DDR RAM at the bottom of this section, and makes note of the fact that the Athlon pretty well matches performance for a better price. I don't think that the authors are biased, maybe they thought the article was encrusted with enough numbers, and trusted readers who wanted to compare to check the relevant review.
This page shows that both Intel and AMD boards have 10-15% memory error rates. This seems pretty high to me -- is this typical? Why?
I hold it, that a little rebellion, now and then, is a good thing. -- Thomas Jefferson
If you would like to see the P4 compared against the Athlon, check this site.
It's as much a 400mhz bus as the Athlon is a 200mhz bus, and I haven't seen anyone complaining about that.
THE YEAR WAS 2081, and everybody was finally equal...
So far from all the reviews I've read, they all say this first generation P4 just isn't worth the buyers effort. First off, the price is enough to scare off most interested folks. Secondly, the benchmarks just haven't been all that great. The chips power comsumption/heat-displacement allows it to fight for the right to be your home heating solution.
My questions would be, now with the new board, will it support the next batch of P4's? One's that can manage SMP? It seams like maybe Asus is jumping the gun just a hair since revised P4's can't be to far around the corner.
Overall the P4 just doesn't add up to be anything but expensive. New chip, new board, new case, new power supply. Thats an aweful lot of new things needed just to get 400Mhz more then the leading compitition. Is it worth it yet? I say no. Things should get very interesting in the next 2 years as we should see the fast approche of the 3Gz mark with offerings from both Intel and AMD and what I think will be the beginning of he desktop server market and the slow end of the desktop pc erra.
Trying to be different, just like everyone else.
Or maybe these guys increased the 400MHz bus to 410 :)
If Intel boss Craig Barrett says Rambus is the wrong tech partner, why is ASUS following up with a RDRAM MB? Ok, it overclocks, but why not using SDRAM or DDR SDRAM? Stunning, really... like rolling out a BMW with a VW engine.
--
A feeling of having made the same mistake before: Deja Foobar
More pedantics. Moore's law does relate to component density, and therefore we really do not expect to see a 2GHZ chip 18 months after the 1GHZ chip. On the other hand, I do think it is reasonable to expect that technology is advancing at least in a compound fashion, so going from 1 GHZ to 2GHZ should be no more difficult that going from 1MHZ to 2 MHZ.
Do you even know what SSE2 is? It's a set of instructions that use special hardware to accelerate very specific operations that are critical to some very computationally intensive algorithms. All of those extensions are the same way.
Someone has to write libraries that use those extensions. Technically, those are what need to be written. Typically, they are written in optimized assembly-level code. Then the applications link into those libraries and call the primitives that use the specific extensions. Alternatively, the compiler will insert sequences of optimized SSE2 code into your application when you include specific macro calls in your source code.
Apparently, Tom's hardware's and Anandtech's self-proclaimed hardware gurus don't quite understand this process. It's misleading when they compare Athlon running applications that were optimized for Athlon extensions against P4 running applications that were optimized for P3. They just don't seem to know better, and I can't really blame them for their ignorance. But they shouldn't claim to be omniscient either.
Apologies for the ObAOL, "me too", but I've been using a deeply old Dell PPro180 with FreeBSD 3.x and 4.x for around a year now. And you know what? It goes like a rocket.
:)
Now admittedly it hasn't been too badly abused, it runs headless for a start and I suspect that helps a lot. I also do a lot of work with networks, so disk IO is neither here nor there.
But I do munch a lot of integer numbers (the processor gets really quite hot), and suspect the memory bandwidth gets caned quite heavily. And also beat on the network cards which, incidentally, are i82559's connected via 33MHz PCI. As opposed to that horrendously old fashioned i82557's connected via 33Mhz PCI. Yay progress!
The point is..... there is no point, this is a me too. The PPro was a fundamentally good chip. It still is, and when you bear in mind the not spectacular advances in (for example) system busses and the still earth-shattering cost of gig ethernet, it's a pretty hard business case to push anything with much more power.
Dave
I write a blog now, you should be afraid.
"Does it seem like the 2nd GHz mark is approaching a lot faster than the first one did?"
It certainly did. Make sure to thank AMD for pushing ahead to 1GHz, and scaring intel into doing a half-assed job of getting there.
*twiddle*
I type corrected. C O R R E C T E D
The 2x increase in performance from 1GHz to 2GHz will be completed in less than 2 years. The industry will accomplish this by rearchitecting their chips to support higher levels of pipelining and by decreasing transistor size to increase transistor speed and lower power consumption.
Russian parts, American parts, all Made in Taiwan!
actually legacy-free, isn't it? Or did they just remove the good ol' ISA slots while maintaining the bridge? I'd hate to have to go through life without having to solve IRQ problems...
People replying to my sig annoy me. That's why I change it all the time.
Wow, one of the reasons I never liked RDRAM was the seemingly universal concept of only two memory module slots; this is just terrible for expandibility. I'm glad to see a board that can go with four, my favorite number of them.
# debian/rules
Wow! and real hair?
I am impressed! 1.68 GHz is 10% faster than 1.5 GHz! Man, that rocks! Woo-hooo!
EagerEyes.org: Visualization and Visual Communication
Well, yes. If power doubles every 18 months, you would expect this kind of increase in growth over time. It took over a decade to get from 4.77 MHz to 100 MHz, and then look what happened.
Apologies if this gets past the pedantry filter...
Does it seem like the 2nd GHz mark is approaching a lot faster than the first one did?
Of course, what with processor speeds increasing exponentially with time (although in this case you're looking at the rather dangerous overclocked speed, rather than the rated speed of the chip)
The P4T includes a few other bits to help with your Pentium 4 experience. There are two RDRAM continuity modules. All four of the RDRAM ports must be filled, and RAM has to be added in pairs. The continuity modules allow you to buy two pieces of RDRAM instead of four. Hey, there must be tradeoffs if you want 3.2GBps of memory bandwidth!
Unless you must fill all of them of course.
People replying to my sig annoy me. That's why I change it all the time.
...but all of my IRQ problems have always come from plug&play on my PCI bus. The damned things always pick their own IRQs and god help you if they choose poorly, or decide to maul your serial ports' IRQs. They LOVE to share IRQs but believe me, contrary to popular belief not that many cards play nice with each other. If you have a geforce and a soundblaster live on the same IRQ for example, you can kiss your system stability goodbye.
Call me old-fashioned, but I preferred it when the cards did as they were told.
---
Where can the word be found, where can the word resound? Not here, there is not enough silence.
"Where shall the word be found, where will the word resound? Not here, there is not enough silence." -T.S. Eliot
Yeah, that's what I heard somewhere else too. Does this only happen when you actually USE the other 2 slots, or all the time?
This is why i love slashdot.
Official GOD FAQ.
Regards,
Tommy
I don't remember ever considering intel boards high performance choices.
Most motherboards reviews generally show intel boards as middle of the road performance wise.
Ok, I'm into more power as much as the next person, but seriously, who needs a personal desktop PC that has clockspeeds of billions of cycles per second? They should be working on how to make it smaller. If they fit a 400Mhz chip in my cell phone, I'd be much happier, than say a Tera-hz machine on my desk.
Although a Thz machine would be pretty nifty.
1) Extremely Realistic Real-Time Rendered fully Immersive VR. Who doesn't want that?
2) Instantaneous Language Translation and realistic vocie synthesis. Who doesn't want a computer that talks back to them with fully customizable inflections and tone? Sexy Female/Male Computer Voice Anyone?
These two apps alone will make for killer games.
Actually most of the above apps become posible at around 20Ghz.
www.enthea.org
1680mhz and 400mhz FSB....oooh...yummmy.....
...mommy can I get one of these for christmas?
.......puleeeeeze?
-Julius X
-Julius X
remove "-whatkindofspamdoyoutakemefor-" from email to send
Intel can't develop it's own DDR or even normal SDRAM motherboard design for months due to a licensing agreement it signed with Rambus.
Asus is using the Intel design for the i850 with a few tweaks, therefore it uses RDRAM.
Light a fire for a man and he'll be warm for a day. Light a man on fire and he'll be warm for the rest of his life.
How come everyone counts performance by mhzs, and/or things mhz paradigm is important to processors? Look at Alpha chips, in their time they did double for each hz, compared to x86 based cpus. Such snippets, as 2Ghz boundary coming up does not really tell me how fast I will compile my code. Really P4 as tested by Toms hardware, says that most software has to be recompiled with new optimizations for SSE2 to take advantage, of P4 so contemplating wether 2ghz boundary is coming up is useless, wether the community and business will support it and majority will be able to take advantage of perhaps ficitious perfomance marker, is the real question.Wasn't slashdot about content and when did it start to change to be website spouting advertisements/advocacy/promotion?
May god help you all *snicker*
Still, 400 mega-transfers per second is fantastic. And it matches the FSB speed (in bits/second) with the speed of dual-channel Rambus. Sweet.
I'm all bug-eyed and pant-y tounged just waiting until a year from now when I can buy the $1999 upgrade for my current P4: P4 Overdrive Kit! Ugrade your current P4 1.4/1.5/1.7 GHz CPU to a P4 Overdrive running at 2 GHz! $1999! Anyone know where I can buy a build-it-yourself home microprocessor kit?
Oooohhh, good one. I see your logic there... that proves how ignorant I truly am.
The article you supposedly read and an excerpt from it. The P4T charged Quake III up to 239.4fps- truly way too fast for anyone. The multiplier was set to 16x and the FSB only worked set up to 105MHz with a 4x multiplier for the RDRAM (420Mhz RDRAM). That was where it knocked itself out. We had to use the 3x multiplier for the RDRAM and thus it wasn't working as well as it could (wasn't quad-pumped).
P.S. I'll just say it for you in case you needed feeding. "YHBT"
I don't like RDRAM. Even tho it runs at a much faster clock speed than SDRAM, the latency is not very good so it ends up being about equal in performance terms. Plus it costs much more than SDRAM, at least in the UK, anyway :)
Besides, the Thunderbird with DDR-SDRAM is IMHO superior to the Pentium (although admittedly in gaming tests the Pentium still wins, but that doesn't matter cos I don't play games anymore), so there!
Check out Tom's Hardware Guide for an article and benchmarks on the first Thunderbird motherboards supporting DDR-SDRAM.
_____________________________________
I've been to Heaven and Hell, and all I got was this lousy sig.
Oops beaten too it.
Official GOD FAQ.
Well, the effort it takes to get to 1GHz is smaller than the effort to get to 2Ghz. Consider the step to 50 or even 100Mhz which I personally found very high at some point. Of course there will be some physical limitations at some point, but apparently the trend will continue for at least a few more years.
;-)
Btw, you probably mean GHz and not MHz
Moz.
see a Text Widget
People keep pointing to Moore's Law every time Intel ups the Mhz bar, without considering that Moore was referring to transistor density on the mask and not clock speed. Anyway, in the time Moore formed his "law" there were no pipelined architectures or multiple bus speeds... everything was one clock tick per instruction fetch and one bus speed.
I note that the change from 33Mhz to 66Mhz and 120Mhz 486 and Pentium systems were primarily the result of clock doubling and tippling, and not main bus speed advances. This has been the trend across the '90s, which is why as CPU clock speed increased performance enhancements per release cycle diminished. Yeaaaa, we're about to break the barrier with 2Ghz systems that are no more than two to three times faster than my PPRO-200. Big deal. Disk I/O is not significantly faster -- though transfer speed from disk cache is great (so what), FSB memory bus speed is a bit faster, and the main I/O and expansion bus is twice as fast (33Mhz to 66Mhz).
This is not a significant advancement, which is why I'm still using my Dual PPRO-200 system quite happily.
--Maynard
Gees, old news, that ASUS has already been benching p4t's with TomsHardware motherboard test of the P4. Tom has already said that the p4t whopps the Intel board bug time. As usual, ASUS come through again, and this time it's an overclockers dream!
Is it just me or are there a whole lot of hoops you have to jump through to use a P4? A new case, a new PS, a FRIGGIN' 1 POUND heatsink, whack-ass heatsink clips, expensive as hell RIMM memory (does this mean that Intel is giving us a RIMM-job?) and in this motherboard's case, a rubber sheet and a second metal motherboard tray?
;-)
Oy! I guess I'll take my P166 anyday.
Quando Omni Flunkus Moritati
Regardless of the specific case Moore was observing, his "law" can be generalized to any number of different technologies. Bandwidth, RAM, storage space, and clockspeed all appear to follow similar curves, though the exact amount of time it takes each to double (six months, 18 months, 2 years, etc.) may be somewhat different, the fact is that "Moore's law" is more than simply a statement about transistor density. It is a general rule of thumb that may be adapted, in some form or another, to fit many developing computer technologies.
"The question of whether a computer can think is no more interesting than that of whether a submarine can swim" -EWD
No, that's Moore's law.
More's law dictates that there will be an exponential number of redundant posts over time on /.
Oh, and alsatian meat is more expensive than mongrol meat.
Is that just some subtile advertising for Intel or did they forget to benchmark it alongside some Athlon, Alpha, bi- or quadri CPUs boards, etc ?
Well all that this article reads is that the newest P4 ASUS motherboard is quicker than the previous...
Well, the opposite would seriously amaze me.
--
Trolling using another account since 2005.