Slashdot Mirror

here's the anandtech review of the board
Goes into quite alot of detail.

metalgeek

No, Rambus is a publicly held company with part ownership by intel.

/* How about profits from licensing Rambus technology */

The same profits AMD is receiving with THEIR license from Rambus. Chipsets define the RAM, not chips. No, Rambus and Intel have no been 100% aboveboard in the way they've approached the market, but Rambus has designed a product with great potential for technical superiority. Anand Tech has two interesting articles discussing the ramifications of RDRAM, DDR SDRAM and SDRAM.

Even the benchmark systems had to use a different setup.

Sorry to rain on your parade. Having said this though, Tom's managed to get a T-Bird 750 to go on a K7V with the latest BIOS, however they couldn't overclock it.

It's probably not what you wanted to hear, but t's all I've been able to dig up. (Having said that it made me feel a little less cheated having bought my 'Classic' Athlon 6 weeks before the T-Bird came out).

Swinging back on-topic, it definitely sounds like the fault isn't with the processor. Most likely either Gateway aren't regulating the voltage too well, or their design's a bit squiffy. I remember that my friend's MSI Athlon system seemed to have a voltage issue, which was sorted by switching to the K7V.

With the right motherboards, the Duron will be a real winner. Maybe stick a HighPoint chip on there, to circumvent Via's and AMD's disk transfer rates which are in the crapper...and give us some overclocking options...and you've got a great opportunity for overclocking heaven if you stick an Alpha on it!

Hopefully Soyo will make a decent Duron board - the 6BA+ IV, their flagship BX model, which my 500E is on, is the best board I've ever used. It's incredibly stable even running 1.5 times faster than normal (image here), and if they make a Duron board I can't wait to see how far people take these things.

Or are there x86 boxes out there that have it?
Even though there aren't many, more and more x86 mobo's have them. Look at anandtech's coverage of Computex, here for some info on some upcoming motherboards with fully-fledged PCI slots. Also, the third day coverage, here has some other motherboards with 64-bit PCI.

On a side note, it is pretty cool that a lot of these newer boards (those supporting ATA-100) have the HighPoint HPT370 controller, which is a RAID controller as well. How convenient! =)

Or are there x86 boxes out there that have it?
Even though there aren't many, more and more x86 mobo's have them. Look at anandtech's coverage of Computex, here for some info on some upcoming motherboards with fully-fledged PCI slots. Also, the third day coverage, here has some other motherboards with 64-bit PCI.

On a side note, it is pretty cool that a lot of these newer boards (those supporting ATA-100) have the HighPoint HPT370 controller, which is a RAID controller as well. How convenient! =)

I've been hoping that someone would bring this up so that I could rehash the discussion we had on RDRAM back when this whole latency story broke. Below you will find a number of links to other places. Suspiciously, out of the holy trinity of hardware review sites (Tom's Hardware, Anandtech, and Sharky Extreme), the ONLY one that speaks up in favor of RDRAM and doesn't talk about its latency problems is Anandtech. Hmm...

From Sharky Extreme on this page:

The memory bus we are all used to operates at 100MHz and is 64-bits wide. Rambus' offering runs at 400MHz (transferring on the rising and falling edges of the clock) and is 16-bits wide. What this essentially translates into is a faster Rambus interface (in terms of frequency) with added latency because of the smaller "width" of the bus.

From Tom's Hardware: This page tells what the theoretical bus bandwidth is for SDRAM, DDR SDRAM, and RDRAM. I quote from the following page:

Continuously managing multiple latencies would be a nightmare for the memory controller. In order to work around this, when a system is booted the RDRAM subsystem performs an involved initialization process to determine what the greatest latency is for the entire RDRAM system and adjusts all RDRAM devices to have the same latency as the slowest RDRAM device on the system. And remember that in a real world system each RIMM will have many RDRAM devices so this latency balancing is quite complex.

(Emphasis is mine.) The next paragraph reads:

An RDRAM chip typically has a normal 20 ns page read access latency. To balance latencies, these chips have a TPARM control register that can be programmed with a 2.5, 5.0, 7.5 or 10.0 ns of artificial compensating latency. This means that the normal chip latency can be as much as 50% higher than the minimal 20 ns often quoted as RDRAM's page read latency. Compare this with the fastest PC100 SDRAM with a latency of only 20 ns, but again remember that RDRAM has even other issues that bring its total latency much higher still.

Finally, An article from Real World Tech explains just what the timings are like, why they occur, and why they mean that DDR SDRAM is going to be faster for the forseeable future. A very instructive paragraph on the general problems with RDRAM follows:

RIMMs also generally require a metallic heat spreader enclosure to avoid an excessive localized heating of any single memory device. Finally, the computer system motherboard into which RIMMs plug must have tightly controlled electrical characteristics that match RIMM circuit cards to avoid unwanted impedance mismatches and signal reflections. This can require extra signal layers and power planes, which along with the tighter manufacturing tolerances, results in a more expensive computer motherboard.

So let's see, RAMBUS memory has higher latency, less bandwidth, consumes more power and therefore dissipates more heat, and it's more expensive. It basically sucks compared to DDR SDRAM in every way... Where's the plus side?

In AnandTech's high-end dream system article they used an Intel OR840 motherboard which is a 2 CPU board with 2 RAMBUS channels.

Anandtech has a very different( and in my opinion saner) take on Rambus part 1 looks at the reasoning behind Rambus while part 2 gives current performance of RDRAM vs. SDRAM and looks at where things are expected to be a few years down the road.

Anandtech has a very different( and in my opinion saner) take on Rambus part 1 looks at the reasoning behind Rambus while part 2 gives current performance of RDRAM vs. SDRAM and looks at where things are expected to be a few years down the road.

Anandtech has a very different( and in my opinion saner) take on Rambus part 1 looks at the reasoning behind Rambus while part 2 gives current performance of RDRAM vs. SDRAM and looks at where things are expected to be a few years down the road.

According to an EBNews article, "...A board using a single VSA-100 chip -- branded under the Voodoo4 name -- is functioning normally."

The V4 (Which has not yet shipped), is basically meant to be a rival for the TNT-Ultra. According to Anandtech's 3dfx Voodoo4 4500 & Voodoo5 5500 Preview, the V4 (with Beta drivers) does pretty good against the TNT-Ultra. It's a little slower at 640x480, a little faster at higher resolutions. At the same time, it offers some extra goodies like 2x FSAA, improved 16-bit color quality, and (possibly) a lower price.

>Is PC-133 "there" yet? Is it a cost-effective improvement over PC-100?

Yes. Assuming your mobo is running at 133, anyway. Via has chipsets out for both intel and AMD processors that support PC-133.

>Is PC-133++ anywhere near? (I.e., PC-150, PX-whatever.)

No. Current PC-133 can often be overclocked to 140-150 Mhz, but that is too near the edge of the performance envelope for commercial use. Faster memory is going to be DDR, by all indications.

>Is DDR memory "there" yet? Cost-effective?

Almost there. It is already in use on graphics cards, as you may know. The technology works. And it is not that much more expensive than normal SDRAM, unlike RDRAM.

> How do you expect the various AMD *ons to shake out in price/performace for a given clock speed?

I would think the order for price/performance will be like this:

Duron, Celeron, "Thunderbird", PIII

The reason is simple: Intel has a serious brand going there (bum BUM ba BUM!), and so AMD need to be that much better to beat them. Right now, AMD is.

>Should I wait a few months? Six months? A year? Or is the price/performance improvement going to be more or less continuous during that time?

Right now, wait. The reason is that right now, the bottleneck in system performance in the PC world is memory bandwidth. Up until fairly recently, this was not a problem -- back when clock multiplier ranged up to 3 or 3.5. But starting with the PIIs and continuing until now, CPU cores are running so fast that they are spending a lot of time waiting on memory.

Go check out the performance of various CPUs in benchmarks, say at Anandtech as in this set of benches of recent processors. What do you see? That at the high end, even with full-speed L2 cache of 256K, PIIIs are still only gaining about 60% of their theoretically possible speedup. That is, a PIII 1000 rates 191 on the SySMark 2000, while the PIII 866 gets 176. 1000/866 is 1.154; that is, the PIII 1G is running 15% faster in core. But it is only getting 8.5% higher performance.

Now go look at this page on Tom's site, where he is exploring the effect of raising the bus speed to 150Mhz. He find some very significant speed ups from raising bus speed; i.e., in Quake III a PIII 975/150 (150Mhz bus) is faster than a PIII 1000/133.

The upshot of this is, that once AMD has a chipset out supporting DDR, they should have a real winner. You might start out just buying a Duron and using old SDRAM in it, but then later upgrading the processor to some mongo Athlon 1.4G or whatever, with DDR SDRAM 133.

Right now the best price-performance you can do is to buy a BX board, some PC-100, and overclock a celeron-II from 566 to 850ish. (Assuming you can find a cel-II; I dunno that you can buy them on the street yet.) Very good price/performance here, but no future.

There is a little information about this at anandtech, which has stuff about the Duron and Thunderbird. Specifically, it says that the Duron is "packaged as a 462 pin socket a" chip, and it even has pretty pictures.
Alas, there is no real information regarding the possibilities of SMP in here. Perhaps AMD has forsaken the multi-processor pc market for the meantime! Woe is me.

If you would like to know why intel chose Rambus instead of DDR-SDRAM, check out this article
and this one.

If you like Tom Pabst better, look at his new review of GeForce2 GTS cards and note what platform he used for the comparison: Intel OR840 motherboard with RDRAM.

If you like Anandtech better, look at his "Dream System": Intel OR840 + SMP CuMine + RDRAM.

(btw, for 1/2 of the $11,000 price he quotes, you could get a Dell Precision Workstation 420 with a better (Nvidia Quadro) video card, and faster processors (866Mhz vs. 733MHz).

So sure, blame Intel for stepping on it's crank multiple times in the past 6 months. But try to understand the technology too.

Here is Anand's Comdex '99 coverage, which is an excellent resource for those wanting to know more about AMD's future.

Also check out Anandtech which has previews of the V4 4500 and V5 V5500.

Also check out Anandtech which has previews of the V4 4500 and V5 V5500.

Anandtech also has a Voodoo 4/5 preview up today. What's interesting is that, yes, at low resolutions, nVidia's GeForce beats it; however, at high resolution (1024x768 and higher) the Voodoo5 catches up and passes the GeForce for a good margin.

High resolution benchmarks often give a good indication of the raw power of the hardware itself. Anand believes the poor perform at low resolution is due to poor drivers, and I'm inclined to agree. As nVidia has shown with the Detonator drivers, it's quite possible that updated versions (like the final ones when it actually comes out) will give the V5 a boost. The important part is all the low resolutions, while slower, are certainly _PLENTY_ of FPS to play with, and, what's more, the V5 makes some of the higher resolutions playable as well.

And the last factor that matters more for Slashdotters... Like 'em or hate 'em, 3dfx has provided traditionally provided very good Linux driver support, unlike some companies (rhymes with binaryonlynoDRIvidia)...

Anandtech also has a Voodoo 4/5 preview up today. What's interesting is that, yes, at low resolutions, nVidia's GeForce beats it; however, at high resolution (1024x768 and higher) the Voodoo5 catches up and passes the GeForce for a good margin.

High resolution benchmarks often give a good indication of the raw power of the hardware itself. Anand believes the poor perform at low resolution is due to poor drivers, and I'm inclined to agree. As nVidia has shown with the Detonator drivers, it's quite possible that updated versions (like the final ones when it actually comes out) will give the V5 a boost. The important part is all the low resolutions, while slower, are certainly _PLENTY_ of FPS to play with, and, what's more, the V5 makes some of the higher resolutions playable as well.

And the last factor that matters more for Slashdotters... Like 'em or hate 'em, 3dfx has provided traditionally provided very good Linux driver support, unlike some companies (rhymes with binaryonlynoDRIvidia)...

Ahh, but even different q3 levels perform differently when comparing certain hardware configs. For example, check out Anandtech article on 64 MB GeForces. The 32 ddr and 64 ddr perform identically with demo 001, but the 64 mb version annihilates the 32 mb one in quaver. The question is, which demo can be considered standard?

Peltiers do generate a lot of heat but, for example, the Swiftech MC2001 uses 2 fans for two thermoelectric coolers (the Peltier junctions), moving 66 cfm of air. Do you think Peltier manufacturers are so grossly incompetent as to heat up the insides of the case? Freezing, that's a different matter. They recommend that you use a silicon compound to prevents moisture penetrating the circuits and then freezing solid..

Anandtech, amongst others, has written a review of Swiftech's MC2000 and MC1000 Peltiers.

I am not, nor have been connected with Swiftech in any way.

____________

____________

• The Tech Report
• Ars Technica
• AnandTech
• HardOCP

The other sites are www.cnet.com and www.sysopt.com if you wondered. :)

Okay, I noticed there's a lot of misinformation/total crap being thrown out on /. right now...so I'll clear a few things up.

Correction: Clock for clock, according to SharkyExtreme's and AnandTech's benchmarks, the Pentium III takes a majority decision against the Athlon while using the i820/RDRAM and KX133/SDRAM chipsets (with the notable exception of professional CAD/CAM), which is useful for the money-is-no-object department. Interestingly, Anand also benched the P3 with a Apollo 133A/SDRAM chipset revealing a give-and-take tie relative to the Athlon, for those of us that are a bit more price conscious.

Correction: The P3 L2 cache is 8-way associative, 256 bit wide, 256KB in size, and runs at full clockspeed. The Athlon L2 cache is 512KB in size, running at 1/3 the clockspeed. The Athlon also has a 128KB L1 cache compared to the P3's 32KB L1 cache, both running at full clockspeed.

Correction: There is NO yield problem at Intel. There is, however, a supply problem, due to management mispredicting what quantity in chips they need to have supplied, as well as reallocation of resources as Intel prepares its fabs for Willamette and Itanium. Gotta love management. For proof, check out the amazing ability of Intel's 500E-600E chips to overclock to 700+ MHz. That's not a characteristic of a chipmaker with yield problems.

Correction: Why on Earth are people deciding what processor is superior by the supply of said chips? Like most sane people, I happen to judge performance on the basis of performance alone. Or maybe it's because I'm not a brand-name zealot. Either way...unless you're talking price/performance (in which case why even talk about GHz processors?) please can the supply arguments.

So who wins? The consumer does. Hopefully with the introduction of Cyrix's Joshua processors, the chipmakers will be squeezed even harder to cut both profits and prices. If you really desire a God Box, go take out a student loan and treat yourself to an SMP Alpha platform.

AnandTech Review of CuMine 1Ghz

This article has "interesting" benchmarks, using the i820 chipset. It shows that the Athlon can beat the crap out of the Intel chip in alot of the cases.

Anandtech has a really good review of the Intel (and AMD) chips. The difference in performance between the two is pretty interesting. While the 1/3 speed cache on the AMD hurts it in some benchmarks, it still whoops up on Intel in others. Very good piece on the technical and business aspects of the speed race.

lots of benchmarks are available here

His claims are far from outrageous. That you look to Sharky for technical insight doesn't rack up any points for your team, either.

If you like benchmarks, you should look here, here, here, or maybe even here.

The Coppermine has a few technical advantages over the Athlon, and even outperforms it on most platforms. The Athlon still suffers from a selection of mostly sub-par motherboards, and Intel's 820 is a dud. There are no great chipsets available to the motherboard manufacturers right now (though Via's KX133 is off to a good start) -- a "wait and see" attitude is probably the best thing for us about now. Poot.

I stand by the statement. Look at Anandtech's latest review of the 1 GHz chip .

I picked the SYSMARK2000 to demonstrate my claim but look at them across the board.

PIII (800) BX PC-100 Ram - 30.8
Athlon 850 AMD-750 - 30.8

Sure, if you enable superbypass (not supported by my K7M motherboard in the revision I bought) or use a Via KX-133 chipset motherboard you can eke out a few more points. The PIII can also get a small jump with RDRAM but it costs a fortune so I'm not using it for comparison.

You busted me. I work for Intel. Read my previous posts.

But, according to this Anandtech article yhey will hopefully be available by the end of 2000.

UNLIKE Serial ATA, 1394 is already supported by Microsoft Windows, and to some extent, Linux.

To quote from the article, "Most important is the fact that Serial ATA is software/register compatible with Parallel ATA, which means there is no need to rewrite anything at the driver or OS level."

Which means that Linux and Windows already support it, and not merely "to some extent".

Firewire is by far the technologically superior standard. I like technologically superior tools; that's why I have two Beta VCRs rotting in my attic. But I think SATA is going to be very popular for internally-connected devices like disk and tape drives, if only so manufacturers can avoid the higher cost of adopting an entirely new standard instead of just building off the existing one. As a manufacturer, which would you rather do: hand the team a full specification and tell them to implement it, or give them your existing code base and say, "hack this so it does I/O one bit at a time instead of eight"?

--

• 2x ALU unit (i.e. the integer processor runs at 3.0 GHz)
• FSB runs at "400 MHz" (similar to the "200 MHz" EV6 bus)
• the introduction of SSE2

EN-8950 review.

Take a look at the GeForce - it does (3) and (4) just like those SGI workstations, and on a $200 (or $300 depending on things like RAM bandwidth) PC 3D card. No, it's not up to $100k SGI workstation standards, but it's getting closer. Nvidia's Quadro (which isn't much more than a souped-up Geforce) isn't a gamer's card, and SGI is working with Nvidia on future PC accelerators IIRC. Take a look at Anandtech's Quadro DDR review and see what you can get on a PC for under $1000.

-jwb

I was very pleased to see that this video card won Video Card of the Year. As of late it's been seriously underated, even after sites like Anandtech declared it the best card for Athlon systems.

Also, Matrox was kind enough to release specs for their G200 (and G400? I think they're very similar in driver writing). This basically makes it the only 32-bit color viable card for Linux gamers.

All of these sites occasionally do exhaustive introductory articles, and even if you can't find exactly what you're looking for, you'll probably still learn a lot from their tangential remarks etc. (kinda like the LDP).

Ars-Technica also has a number of discussion forums that might be able to give you a hand.

I like the Anandtech review as well.

Pablo Nevares, "the freshmaker".

If these are going to be used as phones, why not overclock the processor to 900MHz (or 2.4 GHz!) to match the frequency of cordless phones?

This would settle those occasional comments about CPU frequency interfering with radio frequency and vice versa, since they don't.

AnandTech, Sharky Extreme, Adrenaline Vault, and Ars Technica seem to have screenshots of a new overclocking record every week or two. I believe it currently stands at ~1300MHz -- is 2.4GHz unreasonable to think?

--

All Athlons 700 Mhz and below use .5 multiplier for the L2 cache with respect to CPU speed. The 750 Mhz part (and 800) runs with a .4 multiplier. This means that L2 cache of a 750 runs at 300 Mhz, a significant reduction from 350 Mhz L2 cache of the 700 Mhz. Intel is now including 256KB of cache on die with the P3's which means that L2 cache gets that same boost in Mhz as the processor. The performance lead the athlon's held over earlier P3's (500-600 Mhz range) is evaporating. Still though, sometime early next year AMD will add L2 on die, which should give it a decent boost.

for more information look here:

www.anandtech.com

celerons are hands down the best processors to buy if you're on a budget and don't let anyone tell you any different. you can pick up a 466 for less than US$70 right now and with some cooling it should overclock to 550 easy. if you go with a 366 celeron you should still be able to overclock it 500+ with cooling - intel's yields on celerons are very good, so a lot of the time there is little difference between the slower and faster models besides the 'official' clock rate. i've read about celeron 300As overclocked past 600mhz. performance is usually about the same as a p2/p3 at the same clock rate (though the celerons use a slower bus...)

i'll have to respectfully disagree with the last poster: you'd be crazy to upgrade the 16 meg banshee when you've got a 166 in there. your processor is much farther behind than your video card.

if visit this site on a regular basis you won't go wrong: www.anandtech.com

it's a slashdot-esque site dealing only with the x86 hardware industry (mainly the gaming side of it...). they link to stories/reviews/etc that other good hardware sites publish. there are many nice hardware sites out there, if you look at anandtech you will quickly find some good ones.

this page does a 'weekly cpu price guide' article in which they also recommend which cpus they think are the best to buy: www.sharkyextreme.com

if i were going to buy a computer right now it'd definitely have an abit bp6 mainboard with a pair of overclocked celerons.

Does Intel think we're supposed to be excited about this? By the time they get that processor out the door, AMD will be at 1.2GHz+ with the Athlon. Check out the benchmarks with that absurd and immature RDRAM they're trying to push on us.
Get a clue, Intel.

Anandtech recently conducted a test to evaluate the performance tradeoffs of a higher clock rate and L2 cache speed.

He concluded that while the reduced L2 cache speed DOES affect some things, clock rate is way more important. This suprised me, the L2 cache speed reduction affected it very minimally.

Athlons have steps of 50 MHz, there is no 733 MHz Athlon, it jumps from 700 MHz to 750 MHz. I saw an interesting review here (hope I can remember how to do links!):

Anandtech Review

I do hope they get the L2 cache on die in a hurry, but of course, think of the transistor count after dumping in another 512 kB of memory!

While the Coppermine does narrow the previously large gap between the Pentium III and Athlon, the K7 is still faster. I've read many different comparisons between the two -- and the Coppermine, clock for clock, is definitely showing its age with its Pentium-Pro core. The Athlon has LOTS of room to grow.

Be careful when reading benchmarks, some tests include SSE enhancements without 3dnow! support. Some 3D performance tests use high resolutions to put the bottleneck in the video card and not the cpu (thus decreasing differences). A good reliable source for benchmarks is Anandtech.

AMD announces the xxx Mhz Athlon

Tom's hardware reviews the ______

Transmeta gets another patent

(MS/Dvorak/Whoever) spreading FUD about linux

amd - enough already - we know they can take kryotech's refregeration tech and make the athlon go beyond 800 Mhz

tom's hardware - so what? - plenty of websites out there are dedicated to computer hardware, why do they always post tom's reviews? stop being biased toward anand! :)

transmeta - patience, my friend, patience - stop speculating and wait for their product to come out.

fud - it's all lies - much of it is just misinformation for the misinformed, and dvorak's just in it for the ad revenues... "stop feeding trolls and don't read spam"




_______________________________________________
There is no statute of limitation on stupidity.

GeForce / TNT2Ultra / Voodoo3 Roundup [ Shugashack]
Guillemot GeForce256 3D Prophet Review [Ace's Hardware]
Guillemot GeFroce256 3D Prophet Review [Puissance PC]
nVidia GeForce 256: To Buy or Not to Buy [AnandTech]
Guillemot GeForce256 3D Prophet Review [GA-Source]
nVidia GeForce256 DDR Review [3DGPU]
nVidia GeForce256 DDR Review [Riva Extreme]
nVidia GeForce256 DDR Preview [Thresh's FiringSquad]
nVidia GeForce256 DDR Review [Riva3D]
nVidia GeForce256 DDR Review [Planet Riva]
nVidia GeForce256 DDR Benchmarks [Bjorn3D]
Guillemot GeForce256 3D Prophet Review [CGO]
Guillemot GeForce256 3D Prophet Review [Fast Graphics]
Creative GeForce256 Annihilator Benchmarks [3DHardware]

It is interesting to compare the 32 bit performance of the G400 and the GeForce

640x480 GeForce has a slim lead
1024x768 about even
1600x1200 G400 slapping the GeForce silly

Expendable 1280x960

Q3Test 1.08 1600x1200

both from Anand's review