"Fastest PC in the World" Runs Athlon at 800MHz 174
Errant Knyght writes "Not sure if it is true, but if it is...I want one." The Tom's Hardware writeup seems believable; lots of specs, pictures, even ordering info. KryoTech, the company that makes it, puts a refrigeration unit under the PC case and cools the uP to -36 degrees C before it fires up the rest of the unit. Looks like fun.
ermm... can you say false? (Score:1)
Re:Fast gets slower every few months (Score:2)
http://www.accsdata.com/drffreeze/Dr%20Ffreeze.
He didn't get all THAT much of a speed improvment, and with MANY drawbacks.. Water was condensing on the cooler, and dropping into the oil, right onto the board..
Also check out:
http://www.wizard.com/users/scfoster/public_htm
Looks MUCH more promising..
CISC faster than RISC at the same clock speed???!? (Score:1)
CISC architectures have generally also outperformed RISC architectures at the same clock speed.
Then I'm confused as to why a PowerPC (take your pick as to which one, 604e, G4 or G4) runs photoshop much faster than an identically clocked Pentium (of any variety).
I'm also confused as to why SGI's outperform virtually everything (except maybe for Alpha's) with 250Mhz MIPS chips.
It seems to me that most RISC architectures actually do more per clock cycle than most CISC architectures. They may have fewer instructions, but they can stack more of them in the pipeline at one time
Re:Jeez (Score:1)
Current processor manufacturing technology can go 20% to 40% faster when cooled at about -25dgC. That increase in speed is not linked to the heat dissipation. If you take a processor that dissipates be it a quarter of a watt, and cool it dows to -25dgC, then it will be able to go 20%-40% faster.
You don't HAVE to cool down the Athlon to ludicrously low temperatures. But if you do, it'll go faster. If you cool your G4, it'll go faster.
So, perhaps the PPC works smarter... But last time I checked, it was lagging pretty far behind the Athlon and even the Pentium III in performance, while being more expensive and needing proprietary, expensive components to support it... The G4 isn't a smart choice
Re:For video games, duh... (Score:1)
Here's the deal with gamers. First, the CPU used to get a major hit because the graphics cards offloaded the work to them. Now the cards take more for their part, freeing the cpu, and making things faster. The CPU also tries to fix some of these things with SIMD instructions. Unfortunately, x86's architecture makes SIMD instructions less powerful, and they don't do to much. AltiVec is SIMD, and it should be amazing.
Another lag was the system bus, as it got clogged with to much data and slowed things down. We went to 100mhz, and on socket-7, there was a huge boost. The cpu was fed quicker, cache made a big performance gain again, and so on. Now we have a 200mhz bus, and rising, plus a 100 / 133 mhz ram bus, so we can forget about this lag. The CPU shouldn't be starving.
But sure, the cpu has work to do. It wasn't the only lag, though. And as I said, 50mhz is ~ 5% of a performance gain. Maybe that's not exactly true, as it will change with clock speeds where 50 means nothing. If you look at the benchmarks for the Athlon and P3, its currently accurate. Maybe we get 20-30 % gain with the CPU, which is nice. If the CPU isn't starved, but can't handle the data quick enough, this is very useful. But is the CPU overburndened? I don't believe so, and do you really get a 20-30 % performance gain? I doubt it.
If the CPU starved, then upping its speed shouldn't help. It would crunch a bit faster, but nothing to outstanding. The celeron chips seemingly defy this, as they have small caches that run at clock speed. They also have a slower system bus (66mhz), so the cache is always filled. Increasing the cache's input means they get to work on more data - thus they were starved and you merely feed them quicker. A celeron at the same speed of a Pentium IIx can be equal in gaming, if we neglect SIMDs. I'd reckon if you o/c a cpu with backside cache, like P-II/III or a K7, it wouldn't make as great a gain, just like increasing the system bus for a slot was less useful then a socketted system.
In anycase, I merely just wanted to try to get people to remember you upgrade to kill bottlenecks, not so you can spend money uselessly and for a week have a bigger ego. The $350 could be better spend elsewhere for most potential buyers, and you never know if your next system will work wih there case. To top it off, Quake3 for Linux should have SMP capabilities, so I'd much rather spend my money on a second CPU then on a small little case with a refrigurator in it, if others games do the same.
Again.. the point is, you attack bottlenecks for your needs. I don't see this as a good stratigy because it might not be transferable to a new system. But like I said, for me my problem was IO/clogged bus (P5-200), and I solved it. Not everyone who whoo'd and lusted after the cooler were Quake kiddies...
Re:900 mHZ (Score:1)
*Errrrrrnt!*
The only way to truly eliminate RF from leaking out of a box it to throw $10,000 into TEMPEST grade casing which protects against RF leakage.
The reason someone would want to blow $10,000 on something as goofy as that lies in a method of snooping using a Van Eck Reciever. A Van Eck Reciever is used to collect pulses coming from your computer (or its monitor) and displays them on a screen or in another readable/recordable format.
Obviously, this could be hazardous to national security when every Tom, Dick, and Harry can set up a Van Eck snooper. Right now, this has yet to happen, but when it does, TEMPEST grade cases will be seen everywhere.
The FCC rating simply means that a device will not cause unwanted interference with another device. The Van Eck device looks for that interference, and if you're in range, finds it.
I would suggest the Slashdotters out there perform this test: Take an old monitor (I know you have a few) and place it near a TV. Use a program to put up some colors/lines/shapes on the screen and tune to channel 1, 2, or 3. You should be able to make out a faint outline of what's on the screen next to you.
Cheerio!
Re:CISC faster than RISC at the same clock speed?? (Score:1)
The anonymus coward who calls me an idiot (or maybe he was referring to the CISC is faster guy?) has no proof whatsoever. What can you expect from an anonymuus coward? Why are they even allowed to post?
Core based design apporach... Multiple core based design approach maybe? I dunno, thats how they refer to it... Several cores, in essence, the many different parts, (integer 2x, fpu, altivec.. whatnot.. It mainly to do with the partial separation of them that allows them to do several instructions of different sizes...
First time I've heard someone oppose RISC. Heh.
Full PPC processor lineup: 601, 603, 603e, 604, 604e, 745, 750, 755, 7400.
601 is the G1, the first PPC... 603's were the low end G2's, 604's the high end G2's. 7xx is the G3, 745's are the new low power ones, 750 were the first G3's and 755 are the standard newer ones. 7400 is the mother of all PPC's the G4.
All are RISC. All able to compete (or outo in 2D) with much higher MHz Pentiums.
Although 4 gigaflops is the very highest - the minimum of 1 billion ops is the real measurement I guess. =) Much higher than Intel's 600million max theoretical. Sure that alitvec makes the max ridiculously high - but even the minimum is much higher...
AltiVec is not like MMX or 3dnow mostly for one reason. You rename a few instructions, press recompile in CodeWarrior, and ding it's AltiVec enhanced. Now tell me how to rewrite a whole program for MMX?
brokeninside also demonstrated something else. That SGI MIPS chip being very low MHz outdoes Px's. MHz is no longer a viable measurement except for similar architectures (if at all). And PPC is definately not like the pentium architecture...
This may not matter by 2001... IA-64 vs. G4 and in a few months later G5. IA-64 starting at roughly 800MHz, sources have said. Now what will G4's or G5's be by that time?? If they double in two years... oh 1+ GHz. Geez oh well.
-Curt
Re:Jeez (Score:1)
Re:you thought fans were noisy (Score:2)
Dr. Freeze (Score:1)
The Dr Freeze Page [accsdata.com]
Re:900 mHZ (Score:1)
My old Apple ][+ jammed my favorite station.
i have seen PIII faster than that! (Score:1)
--
http://www.beroute.tzo.com
yeah! (Score:1)
Especially if it could cool the drink in a matter of seconds!
Re:Clock rate... what the bid deal ? ? (Score:1)
Re:Clock rate... what the bid deal ? ? (Score:2)
I agree with you that it seems to be a "cool idea" looking for some sort of clue in order to actually be valuable.
It's doing IDE I/O, which means it's not going to be a "barn-burner" from an I/O perspective; all that it really has is a CPU that will doubtless be outmatched by whatever comes out a year from now, with the serious cost of having to pay for a really serious cooling system.
The market sector I'd see it being "hot" in would be that of computer graphics. That is a sector where the priority is forcibly on CPU power.
I think I'd want to use a high end graphics card ($300+) and 256MB of RAM to let the machine really shine.
I didn't see the previous model (K6-III) as being terribly viable; the Athlon feels "less overpriced," but still somewhat pricey for relatively little value.
The merit may come next year when faster Athlon CPUs can get their speed doubled, thereby providing some more massive performance enhancements. Speeding up one CPU may be of little value, but doubling the speed of an SMP box to provide a couple "Gigahertz" worth of power may well be worth $1K for the cooling system...
.. what a waste (Score:2)
So, you want more performance for your x86? For what? Upgrading the other components will free the CPU. If you go scsi, and even cheap, low end
scsi drives, your CPU gets a boost (UDMA has reduced it, but it still is noticable). SCSI you can carry over to your next upgrade. CPU being eaten by graphics? Hate to tell you, but why spend a few hundred on a cooling unit (where 50mhz on x86 ~ 5% performance gain), instead of a new graphics card?
You can lust after it... but its a waste of money for most people. Now if it kept the system cool too, that would make it a bit more desirable. I can't tell you how painful trying to fix heat problems (from SCSI mostly) can be.. but then again, buying smartly kept my system humming along at acceptable speeds. (and designing your own home-brewed $60 system cooler is fun!)
Two-way Pentium III running at 1 GHz (Score:1)
Results [hardwarecentral.com]
Full text [hardwarecentral.com]
Uh... Kinda old, isn't this? (Score:1)
Not that I wouldn't jump on this in an instant if I had the money. I've always wanted a 50lb refrigeration unit.
Fast gets slower every few months (Score:1)
Wait till they can chill a 1 gig Athlon.
Say, remember that guy that used mineral oil
to cool his motherboard? Any news from him
lately? I think I lost the link to his page.
Unreal? only a little bit... (Score:1)
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
Cover of computer shopper.. (Score:1)
Kryotech's cooling system (Score:3)
yeh, your right (Score:1)
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
Old Hat... (Score:1)
Re:Clock rate... what the bid deal ? ? (Score:1)
Guilty of living in the past ... s'true I've used nothing but SCSI for years ....
I had to get off my fat ass and go look
You have a point here. It is true the new ATA/66 controllers come close to a SCSI controller in CPU usage and efficient bus usage.
As the IDE interface grows up SCSI moves on though. At the moment 80 meg/sec is standard and U3W is now offering 160 meg/sec, quite a bit faster than a PCI bus. This of course will only be useful on that bus (drive to drive, drive to device).
CC
Re:1 GHz? (Score:1)
Re:Noiselevel (Score:1)
900 mHZ (Score:3)
Since most new cordless phones transmit at 900 mHZ, having a processor generating the same frequency would cause problems. When the waves from the phone and processor collide, it could cause each signal to cancel the other out, nullifying them both. So how is this problem solved? 900+ processors will actually be built with shielding. Sounds crazy, but it's true.
A bit offtopic, but just a tidbit of info for everyone.
-- Give him Head? Be a Beacon?
Re:Clock rate... what the bid deal ? ? (Score:1)
This is like saying the sex is overrated. After all a bottle of lotion and a nudie mag will cost less than one Christmas or Birthday with a girlfriend. Right?
SCSI is much better than IDE or EIDE or UDMA33. A nice defragged 7200 RPM SCSI2 drive will pound the snot out of a UDMA 33. Wide SCSI2 under similar circumstances will pound a UDMA 66 into the ground.
At home and at work I have multiple computers some with SCSI, some with IDE there is no comparison between the two. Watch a computer with a PII-300 and an IDE hard drive boot slower than a P-233 with an UWSCSI2 and you'll see what I mean. I run Red Hat 6 on both systems and the P-233 with 32mb boots faster than a PII-300 with 64.
Apps launch faster with SCSI. Disk I/O is so much better with SCSI that it doesn't take a rocket scientist to see that it's faster. If you can afford it SCSI is the way to go.
LK
Kryotech systems (Score:1)
Kryotech Renegade [kryotech.com]
Overclocking the Athlon.... (Score:1)
but check out his Athlon overclockign info, defently the 'old' tom, riping surface mount transistors of a $800 PCB... oh yeh
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
Re:Kryotech's cooling system (Score:1)
Through the radio (Score:1)
For a while I had a program called CPUcool that would turn the CPU off via a hlt instruction when it was idling. This actualy produced a very strong noise, It kept the CPU cool though
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
What in the world do you use you mhz for? (Score:1)
Em, that might be true if you are using it as a server or are in need of some sort of ultra fast performance workstation, but for the only application where those extra 200 mhz are actually needed, what harddrive you have makes shit all difference.
can you say Quake?
I wish I was only kidding here...
-
Re:900 mHZ - You're Confused (Score:1)
So, you have a chip running at 900 MHz, that means the clock is putting out a pulse waveform which looks something like a 900 MHz square wave, with a good number of the transistors changing state with the clock edge. Now, let's do a quick fourier series of our friendly 900 MHz square wave, and what's the base frequency, you know, the frequency with the most energy, looks like 900 MHz to me.
Granted, you're not coupling the clock chip to an antenna, so you're not going to have the most efficient transfer of our friendly 900 MHz electrical signal to an EM (Radio) Wave, but it's still going to happen, and could well mess up other electronics running at 900 MHz if it wasn't for the nice metal case acting as a faraday cage.
Remember, at some level, every digital abstraction layer has an analog underpinning.
Re:Old Hat... (Score:1)
Re:Through the radio (Score:1)
Yes and No (explanations) (Score:1)
Other architectures blur the line between RISC and CISC as well. The Alpha architecture is a very pure RISC design, but the Sparc and PPC designs fall in between: they have fixed length instructions but multiple addressing modes (and multiple registers sets in the Sparc).
Nope. The theoretical max MIPS/MFLOPS of any modern processor is simply equal to the clock speed multiplied by the number of simultaneously executing instructions that can be completed in one clock cycle. It has nothing to do with instruction size. Since the PIII has two integer units and one FP unit, it's theoretical MIPS is 2xMHz and MFLOPS is 1xMHz. For the Athlon, theoretical MIPS will be 3xMHz and MFLOPS is 2xMHz. These figures do not take into account SSE or 3DNow! Comparing MIPS & MFLOPS is even more meaningless (and misleading) than comparing MHz. The PIII, Athlon, and G4 all have SIMD (single instruction, multiple data) floating point instruction units, SSE, 3DNow!, and AltiVec respectively. Although AltiVec is clearly the most powerful of the three, all perform instructions on multiple sets of data at the same time. Theoretically, in certain cases, the AltiVec unit can perform 8 single precision FP operations per clock cycle on 4 32-bit FP numbers with one instruction. Apple counts 8xMHz as their max MFLOPS. But your MFLOPS number for Intel doesn't count SSE. Therefore, you are comparing apples (pun intended) and oranges. If you did count SSE and 3DNow! instructions, the numbers for the PIII and Athlon would be higher, but not as high.The SPEC benchmarks, which do not take into account AltiVec or 3DNow!, show that the Athlon has significantly better integer performance and a little worse FP performance than the G4 at the same clock speed. However, if you compare the top of both lines (500MHz G4 and 700MHz Athlon), the Athlon has *much* better integer performance and a little better FP performance. Although there are no direct comparisons available, AltiVec should be much faster than 3DNow! at similar tasks. Therefore, the G4 ought to excel at signal processing and image manipulation. It is also probably faster at 3D graphics & 3D games. The Athlon will excel at most general business and high end applications.
Default Threshold Reached (Score:1)
First, if you were to look at my computer you'd notice the clock is slow. If my computer were operating at 800 megahertz, I'm assuming it'd be fast, or possibly just not as slow.
Second, no matter how much you argue about which computer is fastest, none will ever compare with Batman's supercomputer. Damn that thing is cool. If only I could play Quake 2 on it, I'd die happy.
Where's the any key?
Intel Pentium III to 993 MHz (Score:1)
Re:1 GHz? (Score:1)
Re:Old Hat... (Score:1)
heat resistant Ziplocs? :) (Score:1)
yeah yeah -1 me allready
thresh!! (Score:1)
granted, they're mostly game reviews but hardware reviews are very thorough and informed. I seem to recall them reviewing this about a month ago.
They did a little linux intro for gamers a couple weeks ago. good shit. nice and anti-microsloth.
www.firingsquad.com
It's slashdot (Score:1)
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
Re:That Refrigeration unit is wasted (Score:1)
That said, if it could cool off a Foster's can like that, work would be a lot more fun.
For video games, duh... (Score:1)
"Subtle mind control? Why do all these HTML buttons say 'Submit' ?"
hmmm can't I do this allready? (Score:1)
Couldn't I just drill a couple holes in my box for cables and dump the case in there????? (Just Kidding Folks)
I'd actually be more interested in cooling off my beer in the PC case
yeah yeah -1 me allready
Re:you thought fans were noisy (Score:2)
Re:Clock rate... what the biG deal ? ? (Score:2)
Re:900 mHZ - You're Confused (Score:1)
Re:Only Windows 98 and NT. What a crime. (Score:2)
Re:900 mHZ (Score:2)
Most people find it funny that I still have a Xenix box, (quad full height 320M HD) on my desk.
Re:Clock rate... what the bid deal ? ? (Score:1)
Because only SCSI can handle the throughput. It's possible to get 120 Mb/Sec sustained with SCSI. That ain't bloody likely with any forseeable form of IDE.
>>Currently, the 10k rpm drives aren't available with IDE interfaces simple because the manufacturers want to maintain a two-tiered pricing structure.
10,000 rpm rives are so fast that it would be a waste of money to make them in IDE.
They could continue to do so if they priced them at the same level as the SCSI drives. If you're right and they performance would be the same, people would pay for it.
>>But the lower speed drives perform equally well in both SCSI and IDE versions.
Older/Slower SCSI devices are on par with IDE. A perfect way to illustrate this is with older Macintoshes. When companies like Sonnet began to release G3 upgrades for Nubus macs one of the Magazines (Mac World/Addict/Or something I don't remember) did a test with MacBench. The SCSI disks on 4 year old NuBus Macs outperformed the IDE disks of last years PCI G3s.
LK
Re:900 mHZ (Score:1)
Re: Dumbed Up for You (Score:1)
Clock Spped is irrelevent when comparing chips of different architectures. (not completely irrelevent, you can compare a g3 to a g3 or a pentium to a pentium, fair enough.)
G4 Clock speed will be up there anyways, so soon they will be defeated in both realms, crappy irrelevent measuring and actual power...
Geez oh well they'll have that beat too.
Is that easier to understand?
G5's will be arriving exactly two years from now specing out 2GHz with a new pipeline and new bus topology, coming in 64 and 32 bit flavors with a newer
Not comparing doubling or suspected improvements there.
Intel will have improved Merceds by that time. Merced is already a large chip, especially with that emulator-sub-processor type thing in it and starting at
Don't bother defending the IA-64. Support another. (K7, motorola's friendly parter's chip maybe?)
Re:900 mHZ (Score:1)
First of all, the present processors (say 450Mhz) already produce interfering signals in the cordless freq. range. Please remember that when we say 450 Mhz clock speed , we are talking about a square wave (not a sine).. aand square waves include really strong magnitude harmonics (infinitely many of them actually but at least the 2nd and 3rd harmonics are of a square wave are considerable strong). BTW, The 2nd and 3rd harmonics of a 450 mhz clock signal are 900Mhz and 1350Mhz.
hmm, coming to the point:
SO, the present processors also have interfering signals but there is no considerable interference be oserved. Why is this??
The outstanding reason is that the motherboard actually does not generate the high clock frequency. The highest frequency signal present on the board is 100Mhz or 133Mhz (correct me if I am wrong.. but even so, that should not be any more than 200 Mhz!)
The 450Mhz signal is generated in the CPU chip itself, by doubling and trippling and quadrpling the motherboard-provided lower freq. clock. (which should be smt. like 33 or 66 Mhz )
So the interfering signals (ie. the harmonics of 450Mhz, or whatever, are generated and stay within the chip). Yes, the power of this clock signal is big, as it serves the whole chip!.
But I suppose the phyisical structure is not very much like one that would transmit this signal in a way to produce considerable interference! Or else we would already be having problems with cordless phones near the desktop..
Bottom line: with the CPU clock rates increasing to the 900 Mhz range, we will not see a terribly big increase in interference with cordless phones or other equipment operating in the those frequencies.. Because the very high rate clock signal will never ever be present on the mother board actually, instead the mother boards will keep providing 33-66 Mhz signals to the CPUs from which they can generate their own clock signals
Definition of a "PC" (Score:2)
Apple used to constantly put out "Macs vs PCs" advertisements, but recently they've changed their tune and they consider Macs PCs.
Apple has been putting out a lot of hype about how their system is the first "PC" to be under export law, but Alphas have been under restricted export law for years.
AMD's Athlon appears to be faster than the G4 from what little benchmarks I've been able to find, but I don't believe the Athlon is under export law, which I find a bit odd.
FOLDOC says a PC is:So I guess being a PC depends on the OS, not the CPU. Does that mean a Windows machine setup for multiple logins is no longer a PC?
Who knows.
--
Re:Refrigerator Unit and Noise? (Score:1)
I've worked with Kryotechs before. (Score:2)
My summer job was with a school system that had just purchased 24 Kryotech systems runing amd k6-3's at 500mhz. (i even got a kryotech t-shirt out of it)
here's my thoughts about kryotech:
They are great to brag about, but they are rarely clocked significantly faster than whats already on the market, or soon to be on the market. Tom himself said that he could run the Athlon K7 at 750mhz with no additional cooling. I don't see the huge advantage to dishing out 1,000$ or more for a barebones systems that will be matched by the CPU makers within months.
On another note: Kryotech told my employers that they would have 1ghz machines out by the end of the year. I'm actually suprised they're not out already considering that tom had no problems clocking Athlons to 750mhz. And yet they supercool the 800mhz chip to -30C (i'm not sure about the Athlons but our machines were running -45C to -47C)
-dox
Never underestimate the power of a small tactical nuclear weapon.
Re:900 mHZ (Score:1)
Hmm...so how come when I was in college, and everyone on the floor had 900Mhz phones, I couldn't tap into their conversation? Where was my interference? And why didn't my 25Mhz Mac screw up my 25Mhz cordless way back in the day? Hmm...Ponderous.
Sure you can! (Score:1)
Re:900 mHZ - You're Confused (Score:1)
If you don't know that electricty induces a magnetic field, you shouldn't be pretending to be an expert
Get a clue.. magnets induce electricity in a wire and electricity induces a magnetic propety (left hand rule) in the wire. Sheesh, i learned this stuff in my introduction to electronics course. (Actually, i knew some of this stuff in Junior High..)
Jeez (Score:2)
The G4 emits 8 watts. It doesn't even need a fan. The PPC just works smarter.
An IBM sales rep used to tell me his favorite sales line when persuading universities to change their crays into AIX boxen (with power processors): Ask them what the cooling cost them every year. Ask them if they could put the saved space into better use.
Having to excessively cool down your computer just isn't very cool.
They sound exactly like pop machines! (Score:1)
My 900MHz phone uses spread spectrum (Score:1)
Re:TOM!! (Score:1)
Condensation. (Score:1)
You know, like when you take a can of Mountain Dew out of the fridge and set it down, and it gets all sweaty?
I saw some Unisys motherboards where they tried cooling the CPU like that. The boards got corroded in about two months from all the moisture and were completely destroyed.
de ja vous (Score:1)
didn't
i think i remember something about that.
if someone has the link, please post it, i'm not lazy - i'm just working
---
Re:1000 mHZ (1Ghz) (Score:1)
I was just looking around, (KryoTech's [kryotech.com] product page) and while I was there, they updated their products page to include a 1Ghz AMD Athlon (K7) machine. It has links to a press release that they gave at a shareholder's meeting, and a Q&A session. Pretty cool stuff. It won't come out until the end of the year, but it will probably beat Intel to market. db48x@yahoo.com
Another note (even though I haven't read the other replies...) 900Mhz cordless phones just have a radio transmitter that is tuned to 900Mhz. Its compleatly analog for all I know. It really has very little to do with creating digital things at the same frequency. (There was an article on the AirPort for Macs, at 2.4 Ghz... same deal. It doesn't actually use a 2400Mhz chip, just a radio transmitter tuned to that freq.)
Re:900 mHZ - You're Confused (Score:1)
-- Give him Head? Be a Beacon?
Re:Clock rate... what the bid deal ? ? (Score:1)
So...? (Score:2)
Other notes of interest:
My Voodoo1 (running at 50MHz) has very little radio shielding. Can't listen to the radio and play quake at the same time...
G4'd still beat the Athlon, I'm fairly sure. I dopn't know everything about them, but unless they have a radically different design aproach from the normal pentium or whatever, they don't touch a g4.
I was explaining this one day to my friend.
G4's use a core-based design approach. In the most basic terms, it's "kinda like a multiprocessor system." I was told by my friend that the normal processor basically does one thing at a time.
And this is pretty much confirmed by some fact sheets on Intel's site explaining how a 600MHz pIII does 600 million floating point ops/sec. (600MillionHertz=600million ops)
The g4 does this differently. Every part of the rpocessor (of course including the floating point unit and AltiVec) can do more than one thing at a time. It has two interger units. That should be self explanitory. It has a 64-bit fpu, than can do 1 64-bit calculation or 2 32-bit and so on... It has 128-bit Altivec which can do 1 128-bit, 2 64-bit, 4 32-bit, 8 16-bit or 16-8 bit. You can then see where Apple got their THEORETICAL max or 4 gigaflops, but it can hold that 1 gigaflop too, which is the big deal.
All PPC instructions are the same size, tradionally. 32-bit. It's part of being a RISC chip, it helps performace some, standardized size(s) of these instructions. A few variants now, but not a big deal, compared to what is normally done on the dark side....
Pentiums and compliant processors do instructions of multiple sizes. Pretty much any size, really, and then they are broken down to smaller sizes... The theoretical max is 600Million for the PIII comes from this. But it cant even hold that, that's the theoretical max (who knows the minimum ops?) while the g4 theoretical max is 4 billion, and minimum 1 billion.
Now I know I went on this tirade abouut pentiums, but frankly, can Athlons be much different? The pentiums are/were a pretty normal CISC architecture. So a 800MHz athlon probably pushes out a very maximum of 900million ops, being extremely generous, most likely 800million. Max.
It can claim to be the fastest PC (personal computer, as opposed to a 1GHz Alpha Monster) by clock cycle alone. Very true. But this does not measure the real potential power of the processor. Theoretical and minimum ops show that.
MegaHertz is an outdated form of speed measurement with things like the core-based design. Basically its like comparing a multiprocessor 500MHz system to a normal 600MHz or 800MHz or whatever single processor system. Are quad PIII 500's or a single 800MHz better? Still can be disputed, depending on what is being run, but one surely has more potential pwoer.
And hey with CodeWarrior already able to do the Altivec stuff, and the libraries out there for months now, they can take advantage of that g4. No problem.
-Curt out.
"Hey there was in eyelash in my nose!?"
"How did you know there was an eyelash in your nose?"
A Round up of hardware sites(Re:Kryotech's co....) (Score:2)
- Sharky Extreme [sharkyextreme.com]
- Ars Technica [ars-technica.com]
- AnandTech [anandtech.com]
- Hard OCP [hardocp.com]
- Ace's Hardware [aceshardware.com]
- CompHardware [comphardware.com]
- Tom's Hardware [tomshardware.com]
- The Tech Zone [thetechzone.com]
- Thresh's FiringSquad [firingsquad.com]
- Review News [reviewnews.com]
I have a few more URL's, but they are only cover specific types of hardware like one for 3D boards, BX mobos, AMD chips, etc. Note these aren't really in any order of preference, but I do read Ars and Sharky the most.--
Re:900 mHZ (Score:3)
Hmm...so how come when I was in college, and everyone on the floor had 900Mhz phones, I couldn't tap into their conversation? Where was my interference? And why didn't my 25Mhz Mac screw up my 25Mhz cordless way back in the day? Hmm...Ponderous.
(a). 900 Mhz phones are digital. This makes it impossible for someone to listen in on them with a $300 Rat Shack scanner. Even if you get the digital transmission and try to play it back, it won't work. 900 Mhz cordless phones use super-light encryption. Once broken, yeah, then you can hear it. And I just have $1000 itching a hole in my pocket for it... (not). And even if you had a 900 Mhz phone, there are at least 25 channels. And even if all 25 channels were used, and your phone decided to stomp on someone's signal, then it would sound like this:
FSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS SSSSSSSSSSSSSSSSSSSSSSSSSSSH.
either that, or a modem like sound. I don't know, and I really couldn't care less...
(b). Your mac was sheilded, since it was FCC approved. RF doesn't leak from sheilded boxes.
(c). If your mac wasn't sheilded (ie... Usually very old, seems old machines got away with FCC ratings, even when their sheilding sucked.) then your phone didn't work at 25 Mhz. There is a very high chance it didn't work at 25 Mhz. I don't have my frequency book handy, but I beleive 25 Mhz is used for RC Planes, boats, etc... I know the 47-49 Mhz or so band is used for old Portable phones, Walkie-Talkies, etc... Public, (d). Take a C64. Turn it on. Turn on a radio to about FM 104.5 Mhz, and set it on top of the disk drive. Access the disk drive (LOAD "*",8,1 [return]). Then come back to me with your revised theory.
(e). For the final insult, take your computer to a local Ham radio operator's place. Take the cover off, turn it on, and leave... This will make them want blood... ;-)
Note: I'm not suggesting RF leakage from computers is horrendous, just that it exists. It is infact, not as big a deal as (non radio operator) people might at first think... I keep my case off, and our TV antenna works just fine, TYVM. :-)
Now if we could just get a toilet. (Score:1)
Re:Definition of a "PC" (Score:1)
Re:Clock rate... what the bid deal ? ? (Score:1)
I couldn't really care less about how fast the transfer rate of a single hard drive is - that's really not the question to be asking when it comes to IDE or SCSI.
The question to be asking is, "Do I expect to ever be asking for more than one hard drive access at a time?"
Admittedly, I'm definitely a power user - I do a lot of audio and graphics, work, 3-D raytracing, that sort of stuff. But my set of four "puny" little 2 and 4GB SCSI-2 Wide drives, in actual use, will kick the tar out of any single or combination of IDE drives on the market today.
See, the thing is, a single hard drive can't transfer at anything *close to UltraATA/66. Only the fastest two or three drives on the market come anything close even to UltraATA/33. What matters is, with SCSI, all the devices on the chain can share the bus! That means on my system, SCSI ID 0 can be loading WinDoze DLL's, ID 1 can be swapping memory pages, and ID 2 can be doing the Fourier transforms on a 65MB audio file... so what that each drive is only transferring at 8MB/s! On an IDE system, only one device can access the bus at a time, so... everybody else waits. That's one of the big reasons that everybody tells you not to have a CD drive on the same bus as a hard drive; the drive will be sitting around waiting for that darned slow CD.
Re:Clock rate... what the bid deal ? ? (Score:1)
The reason SCSI is quite a bit more expensive that IDE is that the SCSI controller on the card is a smart hunk of silicon.
The IDE controller on the mobo (nearly always) is a brainless thing and uses CPU cycles to do it's IO.
The SCSI controller has it's brain on board and all the system has to do is hand off it's IO to the SCSI controller which will take care of it with very little CPU usage needed.
DMA enabled or disabled has nothing to do with this. The data gotta' be writ and red.
DMA is really just a way to get around the software memory manager which will always be slow.
Just ask yourself why *every* serious server runs SCSI if what you say is true.
CC
Noiselevel (Score:1)
HC gamers suck regardless (Score:1)
They ABSOLUTELY HAVE TO HAVE the latest and greatest of "x" so they can impress their other weiner friends and everyone else who listens.
I know the type.
Personally, I couldn't care less whether I can play Quake at 1200x1024 because I overclocked my CPU, whereas before I could *only* play at 800x600.
As for the above poster, have you played Unreal? I hate waiting for the levels to load...
PPoE
The G4 excels at (Score:1)
As I understand it, G4's integer performance does not vary greatly from the G3 (at least not more than what you would expect from a 400-500mhz jump). It's the FPU performance that's off the scale, particularly when applications are written to specifically take advantage of the 128-bit vector unit (Velocity Engine/AltiVec).
Things like media encoding, SETI@Home, Photoshop, scientic modeling simulation programs, anything with heavy number crunching really flies on G4s. Of particular interest is the fact that the G4 is quite good at the type of calculations required for cryptography.
Additionally, 7400 (G4) requires little power, and produces very little heat, which means it should have no trouble transitioning to PowerBooks relatively soon. No supercooling required!
- Scott
------
Scott Stevenson
Re:Clock rate... what the bid deal ? ? (Score:1)
The real reason to use SCSI is of course because an IDE controller will use about 30% of your CPU to do it's work.
A SCSI system will use on the order of 2% CPU to do the same thing.
For a server this is importent for sure, but also for yer high end gaming system too.
That's where 2% CPU will save yer ass!
CC
Re:Definition of a "PC" (Score:1)
The thing that determines whether you're under export control or not is how many operations per second you can do - no, not FLOPS, any operation. The AltaVec is very, very good at doing a particular kind of operation very, very quickly - thus putting the CPU as a whole over the limit.
Re:Through the radio (Score:2)
Re:You mean "multiple core" (Score:1)
G4's use a core-based design approach
I believe you mean "multiple core design". All CPUs have something that passes as a core.
- Scott
------
Scott Stevenson
The answers are obvious (Score:1)
The answer is obvious. AMD is secretly producing these things for space aliens. And when the humans discover their base on Yuggoth, they will have the computing capacity to crush all resistance in their path.
On the other hand, it could be because some nimrod just decided to take 'cool computing' too literally.
Besides, hey, I'd pay $2,200 for a PC that looks like a steam-powered marital aid.
Re:I've worked with Kryotechs before. (Score:1)
I would also assume that because of the better cooling, the systems would be more stable than standard air-cooled ones that have been overclocked.
Does anyone have some stability measures (as opposed to anecdotal evidence) of cryo-cooled systems vs. air-cooled?
Cool Alphas? (Score:1)
Re:900 mHZ (Score:4)
TOM!! (Score:2)
HEY TOM! Wake up - there's alot of us out here at work that can't bypass our firewalls easily. You want us to all view your pretty advertisements right? Stop meddling with the http referer then - it's not a mandatory part of the HTTP protocol! Fooooooo....
--
User upgradable (Score:2)
Re:900 mHZ (Score:3)
Clock rate... what the bid deal ? ? (Score:3)
Am i missing something here or has INTEL marketing done such a good job that people who should be thinking of ways to improve performance in other ways are just taking part in the rat race for more clocks ? ?
EIDE, now thats something that i find funny, if you are going to do soooo much to a system, at least have the decency to put in a nice Ultra2Wide drive in there. I dont care what numbers say, SCSI still makes a lot of difference!
Secondly, how about some more memory please ? ?
then comes the question that is it really worth the darn hastle ? things going to be slow in a few months anyway, why waste money on a few clocks if you can spend it wisely on otherthings ?
Dont get me wrong, in practice its cool to see what a processor can push, but we have seen it too many times. Its bad enough that all the marketing hype is about "xxxMhz".... sure, lets put a 600Mhz, Celery, with slow RAM, a crapy vdo card, slow Drive, and call it top of the line!
On most things a system with well chosen components will outperform (in the real world) one that has just a faster processor!!! why we keep getting these "supercool" things is a mystery to me.....
I am not trying to start a war here, i am just saying that maybe articles like this should include a fair warning to users about alternatives to FASTER processors!
just my
Re:Through the radio (Score:2)
Granted, video cards *can* saturate the PCI bus and starve your soundcard of the bandwidth it needs, but this kind of interference will come through even if no sound is playing.
Not quite the fastest in the world... (Score:3)
Oh yeah, it's true, but (Score:3)
Tom covered this back in early August, when he was in yet another "I'm better than you, puny humans" phase - he wrote an article announcing that he had figured out how to overclock Athlons, but that he wasn't going to tell anybody becouse it was just too technical for mere mortals. After a suitable interval of being flamed by the universe, he spilled the beans a couple of days later.
Tom's has gotten a little less interesting of late. I think part of that is because the hardware scene has been relatively static lately, - Athlon is the first radically new thing in a while as far as motherboards/processors go in the Wintel space (I'm not counting G4, guys - Tom doesn't cover that!).
- -Josh Turiel
You should see it running with some decent video.. (Score:2)
That Refrigeration unit is wasted (Score:2)
I am pretty sure that the small fridge on the bottom could easily handle more than that. KryoTech should come up with Renegade Deluxe case that has a round slot (preferrably motorized) on front where one could insert a 32oz can of beverage of choice (Dew, Jolt, beer) for cooling. It would also include cross-platform Cool Alerter (TM) software that would monitor the temperature of the drink and notify you when the drink is ready.
Now, *that's* the case I would buy!
Re:900 mHZ (Score:2)
There was a recent Slashdot article [slashdot.org] about deliberately generating a transmission with your system, but apparently it isn't simple!