[Rant] Pentium IV Prototype at 3.5 GHz | ANN.lu |
Posted on 29-Aug-2001 09:05 GMT by Christian Kemp | 41 comments View flat View list |
Solar (BAUD) wrote:
PPC vs x86, Intel vs AMD, believe me folks I´d like to see PPC rule the market, but this looks like x86 is here to stay. I think it´s interesting to see what will be tomorrow, even if it has little to do with Amiga today.
Yes, I know that GHz isn´t everything, but that´s a *lot* of GHz to make up for with superior (?) architecture...
From Heise Online:
"In the early morning, shortly after 8 AM local time, [Intel] presented a prototype of a 0.13µm Pentium-4 (Northwood), which - shortly - ran at 3.5 GHz.
On average, the IDF record breaking CPUs reach the market after roughly a year. The 2 GHz type that started production yesterday was also presented as a prototype on last year´s IDF. So we can expect the 3.5 GHz Pentium 4 in late summer 2002. In general, so Otellini, the architecture can be scaled up to 10 GHz."
|
|
Pentium IV Prototype at 3.5 GHz : Comment 1 of 41 | ANN.lu |
Posted by Anonymous on 28-Aug-2001 22:00 GMT | Yes, and therefore we need faster and cheaper PPC - Apple should kick Motorola's ass real hard. |
|
Pentium IV Prototype at 3.5 GHz : Comment 2 of 41 | ANN.lu |
Posted by Hans-Joerg Frieden on 28-Aug-2001 22:00 GMT | Let's first see when this thing comes out, what the price will be, and most importantly how many stages they had to add to the pipeline again to get this increase in clock frequency.
They will probably have 30 or more pipeline stages now, and will again invent a nice buzzword term like "GigaPipelined Architecture" to cover the fact that any stall in their pipeline will make their CPU perform 20 times slower :-)
Apart from that, Motorola and IBM aren't idle, either. |
|
Pentium IV Prototype at 3.5 GHz : Comment 3 of 41 | ANN.lu |
|
Pentium IV Prototype at 3.5 GHz : Comment 4 of 41 | ANN.lu |
Posted by Tinman on 28-Aug-2001 22:00 GMT | That's an obscene frequency. In a few years, the number of mobile phones, multi-GHz processors and other futuristic crap is gonna turn the planet into a giant microwave oven :)
I'd like to see what Intel spindoctors reasons are for us needing processors that fast :) |
|
Pentium IV Prototype at 3.5 GHz : Comment 5 of 41 | ANN.lu |
Posted by koan on 28-Aug-2001 22:00 GMT | While x86 based PCs are getting faster and faster, who is using these
extra processing cycles ? Anyone ?
The damn things still take minutes to boot (and reboot!)
You need fast data transfers e.g. to disk and back for the machine to
"feel snappy", which is what users really want at the end of the
day.
Why are Amiga users so dedicated to finding a "PC slaying machine" ?
It may be possible but why bother ? Do they really think millions of
PC users are suddenly going to say "shoot, I've been wrong all
this time" and go off and buy some new turbo Amiga when it comes out ?
I don't think so. Let's focus on doing it right. (But if AI gets a move on
that would be more preferable). |
|
Pentium IV Prototype at 3.5 GHz : Comment 6 of 41 | ANN.lu |
Posted by Solar (BAUD) on 28-Aug-2001 22:00 GMT | In reply to Comment 5 (koan): > While x86 based PCs are getting faster and faster, who is using these
> extra processing cycles ? Anyone ?
> The damn things still take minutes to boot (and reboot!)
1) We said almost the same back when the x86 CPU family became faster and faster while our 68k couldn´t keep up. Now it´s PPC and x86...
2) What takes minutes to boot isn´t the CPU, and it´s not even Windows: It´s the loads and loads of utilities, drivers, tools etc. (e.g. RealPlayer) that are started automatically. I´ve seen Amigas with a loaded WBStartup taking minutes to boot, and I´ve seen bare-bones Windows PCs booting in a couple of seconds.
> You need fast data transfers e.g. to disk and back for the machine to
> "feel snappy", which is what users really want at the end of the
> day.
Data transfers don´t come any faster than ATA-66. (Yes, there is ATA-100, but hard drives have yet to reach 66 MByte/sec. - that´s again independant of CPU *and* OS...)
> Why are Amiga users so dedicated to finding a "PC slaying machine" ?
Because a new Amiga has to be "PC slaying" in order to get out of it´s niche and survive. There´s simply no way to do so hardware wise, so we have to rely on a "PC slaying" OS. Since the hardware doesn´t matter, I think it´s worthwile to look where the x86 CPUs have gotten.
BTW, by the time the PPCs achieve comparable speeds, *their* pipeline won´t be much shorter... |
|
Pentium IV Prototype at 3.5 GHz : Comment 7 of 41 | ANN.lu |
Posted by Graham on 28-Aug-2001 22:00 GMT | In reply to Comment 2 (Hans-Joerg Frieden): The P4 has 20 pipeline stages. The 3.5GHz P4 they showed off will have 20 pipeline stages. The shrink to 0.13u, plus some killer cooling system will have been the reason that the demo CPU got to 3.5GHz.
Anyway, 1 year is 2/3rd of a Moore's Law period (yes, I know it is about transistors, not clock speed, but in the end it has been roughly the same). Today: 2GHz. 2/3rd of 2 is 1.333, which makes a speed of 3.333GHz in 1 year's time anyway. Nothing special there.
Intel are going for reduced IPC and higher speeds in a cyncical and consumer-tricking move. AMD are going for IPC - the Clawhammer apparently has 50% more IPC than the Athlon! If a 1.4GHz Athlon can match a 2GHz P4, then a 2GHz Clawhammer will more than match a 3.5GHz P4.
The Athlon will only be at 1.6GHz at the end of this year, when Intel will be at 2.2GHz. The G4+ will be at 1GHz. Lets hope that next year the PPC will gain some massive speed gains with the G5, although 2GHz looks to be the limit for next year already.
The fact is, to get huge speeds, you need more pipeline stages. However, speed doesn't mean power, as the P4 has demonstrated by having an appalling IPC compared to the Athlon, the PIII and the G3 and G4. |
|
Pentium IV Prototype at 3.5 GHz : Comment 8 of 41 | ANN.lu |
Posted by on 28-Aug-2001 22:00 GMT | In reply to Comment 7 (Graham): >The Athlon will only be at 1.6GHz at the end of this year, when Intel will be >at 2.2GHz. The G4+ will be at 1GHz. Lets hope that next year the PPC will gain >some massive speed gains with the G5, although 2GHz looks to be the limit for >next year already.
Mmmmm, a little bit hopeful. All the tests I've seen suggest that a 1.4G Athlon is roughly comparable to a 1.7G Pentium (socket 423). The Athlon convincingly wins some tests, while the Pentium wins others. I would have thought that a 2G Pentium (again, socket 423) would have a slight advantage. Reason I say socket 423 is that they use the Intel 850 chipset and so use rambus memory which has a bandwidth at the lower end of DDR DRAM (which the 1.4G Athlon uses). It has been suggested that the Athlon is only able to be comparable at the mo because it can achieve higher data rates to and from memory. The new P4's (socket 478) are utter shit, as they use Intels 845 chipset, so using SDR DRAM. Tests have shown that 2G pentiums (478) are no better whatsoever than the Athlons. They are simply crippled with bad memory transfer. But in 6 months time, there'll be another revision of the Pentium which'll use DDR DRAM. Coupled with frequencies that are supposedly going to be starting around the 2.2 to 2.4G, you may see a wider margin appearing between the Pentium and Athlon, irrespective of the 20 pipeline stages.
To the question earlier of why all the GHz, well, the GHz is a great marketing ploy. Intel have got competition so they don't rest on their laurels and continue to try to push up the numbers. Joe Public laps it up cos the bigger number chip is perceived as being better, and given the choice, he wants the better one. If Intel didn't push this way and the Athlon came up with a 3GHz chip with the Pentium stuck at 2Ghz, Intel would start to lose sales big time.
As for who would use this number of cycles, in the greatest numbers it would be the game player. Increase in CPU cycles can increase framerate (as the P4 tests have demonstrated in the main). |
|
Pentium IV Prototype at 3.5 GHz : Comment 9 of 41 | ANN.lu |
Posted by nOw2 on 28-Aug-2001 22:00 GMT | In reply to Comment 6 (Solar (BAUD)): > Data transfers don´t come any faster than ATA-66. (Yes, there is ATA-100, but
Bah. I piss on your IDE.
Or at least, SCSI does. |
|
Pentium IV Prototype at 3.5 GHz : Comment 10 of 41 | ANN.lu |
Posted by Amifan on 28-Aug-2001 22:00 GMT | In reply to Comment 7 (Graham): Unfortunatly, you can call the average PC owner dumb. They think that more Mhzmore speed. So Intel got the advantage if they have a higher Ghz CPU even if an AMD with only half the clockfrequency is faster.
This is how it goes today and how it's going to be in the future.
So it would be better for Motorola to build a G4 with a 200 stage pipeline running at 6 Ghz but with the speed of een G4 running at 1Ghz then the other way around..... |
|
Pentium IV Prototype at 3.5 GHz : Comment 11 of 41 | ANN.lu |
Posted by boingdude on 28-Aug-2001 22:00 GMT | What to do with all those CPU cycles? Amithlon would surely benefit from it greatly. |
|
Pentium IV Prototype at 3.5 GHz : Comment 12 of 41 | ANN.lu |
Posted by Nick on 28-Aug-2001 22:00 GMT | In reply to Comment 11 (boingdude): Not to mention all the 3D rendering software, CAD, compilers... |
|
Pentium IV Prototype at 3.5 GHz : Comment 13 of 41 | ANN.lu |
Posted by Graham on 28-Aug-2001 22:00 GMT | In reply to Comment 8 (): <<All the tests I've seen suggest that a 1.4G Athlon is roughly comparable to a 1.7G Pentium (socket 423).>>
The i850 chipset also supports Socket 478, btw. The tests I have seen put a 1.4GHz Athlon at about the speed of a P4 2GHz on i850 with socket 478. The P4 wins media benchmarks, Quake III and memory bandwidth, the Athlon everything else, for 1/5th the price.
<<Reason I say socket 423 is that they use the Intel 850 chipset and so use rambus memory which has a bandwidth at the lower end of DDR DRAM (which the 1.4G Athlon uses).>>
i850 uses dual channel RDRAM to get a high-latency 3.2GB/s bandwidth - much more than a single channel DDR bandwidth of 2.1GB/s (2.7GB/s soon). DDR has much lower latency than RDRAM though.
<<It has been suggested that the Athlon is only able to be comparable at the mo because it can achieve higher data rates to and from memory.>>
No it hasn't. It can't. It competes because of its superior architecture, albeit an architecture designed for 500MHz -> 4GHz over its lifetime, not 1.4GHz -> 10GHz. In fact, the opposite is true, the P4 compares to the Athlon only because of its huge bandwidth. AMD will release the longer pipelined Clawhammer and Sledgehammer processors next year, which will push performance really high. Intel must be worried about that, hence their MHz push at the moment. Shame that the market is in a slump, eh?
<<The new P4's (socket 478) are utter shit, as they use Intels 845 chipset, so using SDR DRAM.>>
Wrong. i850 supports both S423 and S478. i845 supports both S423 and S478. The new socket is required for higher speeds, as more power needs to be supplied to the processor. It is also about the best thing about the P4, a dinky cute socket.
<<Tests have shown that 2G pentiums (478) are no better whatsoever than the Athlons.>>
Tests have shown that a 1.8GHz P4 on i845 performs like a 1.2GHz Athlon if it is lucky. P4 + SDRAMcrap.
<<But in 6 months time, there'll be another revision of the Pentium which'll use DDR DRAM.>>
You mean now, don't you? And it isn't the processor that supports RDRAM or DDRRAM, it is the chipset. VIA P4X266 supports PC2100 DDR RAM. i845DDR only supports PC1600... SiS645 supports PC2700 DDR RAM!
<<Coupled with frequencies that are supposedly going to be starting around the 2.2 to 2.4G, you may see a wider margin appearing between the Pentium and Athlon, irrespective of the 20 pipeline stages. >>
Yes, and AMD will have 1.53GHz Palomino Athlons out in the first week of October, which will be equivalent to a 2.2GHz P4. Palomino supports SSE and prefetch, hence it will have even better performance.
<<To the question earlier of why all the GHz, well, the GHz is a great marketing ploy.>>
I will agree here.
<<As for who would use this number of cycles, in the greatest numbers it would be the game player. Increase in CPU cycles can increase framerate (as the P4 tests have demonstrated in the main). >>
Increase in memory bandwidth can increase framerate. Usage of SSE and SSE2 instructions can increase framerate. Increase of MHz when using the same processor can increase framerate.
AMD seem to be going towards a PR (P4R?) style system for their processors however. Don't be surprised to see the 1.533GHz Athlon being sold as the "Athlon 2200"...
Graham |
|
Pentium IV Prototype at 3.5 GHz : Comment 14 of 41 | ANN.lu |
Posted by Hans-Joerg Frieden on 28-Aug-2001 22:00 GMT | In reply to Comment 6 (Solar (BAUD)): I don't know the number off my head, but I am sure the PPC still has less than 7 or so pipeline stages. |
|
Pentium IV Prototype at 3.5 GHz : Comment 15 of 41 | ANN.lu |
|
Pentium IV Prototype at 3.5 GHz : Comment 16 of 41 | ANN.lu |
Posted by Ian Shurmer on 28-Aug-2001 22:00 GMT | In reply to Comment 6 (Solar (BAUD)): > "I´ve seen Amigas with a loaded WBStartup taking minutes to boot, and I´ve
> seen bare-bones Windows PCs booting in a couple of seconds."
You're joking aren't u?! A couple of seconds...?!
Ian |
|
Pentium IV Prototype at 3.5 GHz : Comment 17 of 41 | ANN.lu |
Posted by Christian Kemp on 28-Aug-2001 22:00 GMT | In reply to Comment 16 (Ian Shurmer): > You're joking aren't u?! A couple of seconds...?!
If you count the hibernation feature of my notebook, then I guess I can say that I can boot to w2000 in just a few seconds. |
|
Pentium IV Prototype at 3.5 GHz : Comment 18 of 41 | ANN.lu |
Posted by Ian Shurmer on 28-Aug-2001 22:00 GMT | In reply to Comment 17 (Christian Kemp): Now thats just cheating!! ;)
Ian |
|
Pentium IV Prototype at 3.5 GHz : Comment 19 of 41 | ANN.lu |
Posted by Bart Vanhaeren on 28-Aug-2001 22:00 GMT | Some real life facts:
I bought a Apple G4 dual 533Mhz machine a few weeks ago. Since I'm participating in the Distributed.Net Bovine project (RSA RC5-64 encryption cracking contest) I downloaded the Mac G4 client to crack keys on my machine. Benchmarking reveals 4.8 milion keys/sec on a single G4 533Mhz (I have two of these babies in the Mac!!). The same test on an AMD K7 Athlon Thunderbird 1.3 Ghz was not better than 4 milion keys/sec.
Not bad considering the G4 has about 800Mhz less cycles to burn and still faster (OK, G4 code is heavely optimised, but so is x86 code on Athlon)
To put things in perspective, my 68060 Amiga at 50Mhz squeezes out 117 thousand keys/second, roughly the performance of an Intel Pentium with almost twiche as much Mhz.
Altough it's a marketing slogan of Apple at the moment, there's a lot of truth in the "MegaHertz Myth". As long as Motorola catches up at about half of the speed of competitive x86 products, there will be no performance gap. To bad the entire market indeed believes in the Mhz Myth... |
|
Pentium IV Prototype at 3.5 GHz : Comment 20 of 41 | ANN.lu |
Posted by Richie on 29-Aug-2001 22:00 GMT | WHAT!?
"this looks like x86 is here to stay"
Duh!
There are actually two companies reseaching and selling lots of X86.
The question is "Is the ppc here to stay or is moto fixing to drop it like a hot potato. |
|
Pentium IV Prototype at 3.5 GHz : Comment 21 of 41 | ANN.lu |
Posted by sutro on 29-Aug-2001 22:00 GMT | In reply to Comment 5 (koan): > You need fast data transfers e.g. to disk and back for the machine to
"feel snappy", which is what users really want at the end of the
day.
>
Couldn't agree more. That is why Sun's 64bit UltraSparc is a better platform than x86. I would add that x86 is a worse platform than PowerPC in almost all regards - except marketing of course (which brings down the price).
Don't think about making a mistake in "upgrading" a good 600Mhz Athlon-PIII system to a P4/Athlon one that rates above 1GHz. Unless you really like games or want to make Intel/AMD rich... Buy a good case, a better hard disk, another peripheral instead. |
|
Pentium IV Prototype at 3.5 GHz : Comment 22 of 41 | ANN.lu |
Posted by Luca on 29-Aug-2001 22:00 GMT | In reply to Comment 7 (Graham): >The Athlon will only be at 1.6GHz at the end of this year,
>when Intel will be at 2.2GHz
So? The Athlon @ 1.3Ghz is faster than the P4 @ 1.7Ghz |
|
Pentium IV Prototype at 3.5 GHz : Comment 23 of 41 | ANN.lu |
Posted by nian on 29-Aug-2001 22:00 GMT | Hrm everyone talks about the greater frame rate in games with a faster cpu, but the real question is why do we need more fps? how many frames can the human eye see? how big does the resolution have to be? 1600x1200 is great imho, all that there is left is the polygons and light sources but thats got nothing to do with the cpu thats only what 3d gfx card you have. SO I ask you, for a home / game user what do you need the mhz for? The only reason would be if companies keep releasing games like "Oni" that have an equivalent 3d engine to Quake2 but needs 700mhz+ to be playable where as Q2 only needs 200mhz. Poor code! Thats what you get with more mhz, not more performance. |
|
Pentium IV Prototype at 3.5 GHz : Comment 24 of 41 | ANN.lu |
Posted by David Shipman on 29-Aug-2001 22:00 GMT | In reply to Comment 9 (nOw2): Well, no it doesn't; UW-SCSI is on an even par with ATA-100 (and IDE-RAID is cheap and slays SCSI (with the exception of SCSI-RAID, of course)) in terms of raw data throughput; certainly I'm aware of the advantages, but I also know I can get an 80gig ATA100 drive for the cost of an 8-gig UW-SCSI drive
Unless I'm charging a lot of people a lot of money to store their data on my server, I can't justify the expense of SCSI... and don't except the (potential) Amiga-buying public to see it any other way... |
|
Pentium IV Prototype at 3.5 GHz : Comment 25 of 41 | ANN.lu |
Posted by David Shipman on 29-Aug-2001 22:00 GMT | In reply to Comment 23 (nian): The reason people talk about high frame rates for games like Quake3 etc is because they are 'real-world' benchmarks; things everything uses, and know what they mean. Is there any point playing Q3 at 200fps? No, but it means when 'the next big thing' comes along you'll have a better chance of meeting the requirements than someone at 50fps...
Is this _important_? No
Should people think this way? Probably not...
But does it happen, is it the way the world works? Well, yes...it is.
Deal with it |
|
Pentium IV Prototype at 3.5 GHz : Comment 26 of 41 | ANN.lu |
Posted by Mike Pearson on 29-Aug-2001 22:00 GMT | In reply to Comment 19 (Bart Vanhaeren): As to speed, as I recall, on that series of motorola chips (060) and
earlier, that this was the external speed and that to catch up the '86
series reported the internal clock speed which was double clocked
hence a Pentium 66 was really a 33 if reportred the same as motorola.
Hence the Amiga 2000-33 (030) was the equivalent as the P-66. As we
had both at the same time, the amiga was actually faster until the 133
came out. Remember, Back 9in those days, Motorola was touting MIPS,
not mz. What ever happened to mips anyway? (I think there were too
many ways to test-report them.) |
|
Pentium IV Prototype at 3.5 GHz : Comment 27 of 41 | ANN.lu |
Posted by Mike Pearson on 29-Aug-2001 22:00 GMT | In reply to Comment 26 (Mike Pearson): I forgot.... the reason I made the prior comment is I wonder if the different
Ways Intel and Motorola report chip speeds is still opperative and hence
could explain some of the discrepancies between Mhz speeds and actual performance? |
|
Pentium IV Prototype at 3.5 GHz : Comment 28 of 41 | ANN.lu |
Posted by Solar (BAUD) on 29-Aug-2001 22:00 GMT | >> "I´ve seen Amigas with a loaded WBStartup taking minutes to boot, and I´ve
>> seen bare-bones Windows PCs booting in a couple of seconds."
> You're joking aren't u?! A couple of seconds...?!
No. If you count off the HW initialisation, which isn´t a factor of CPU or OS - AmigaOne´s will have to run through that just the same. A bare-bones Win98SE takes less than 20 seconds to boot, at least on my machine - haven´t had the opportunity (or time to kill) to do throughout testing on this.
If CPU would be any factor here, LinuxPPC would have to be faster-booting than Linux-x86. Is that the case?
About the IDE vs. SCSI issue:
ATA-33 -> 33 MB / sec.
ATA-66 -> 66 MB / sec.
ATA-100 -> 100 MB / sec.
U-SCSI -> 20 MB / sec.
UW-SCSI -> 40 MB / sec.
UW2-SCSI -> 80 MB / sec.
SCSI-160 -> 160 MB / sec.
SCSI drives are either none the faster than IDE drives, or if they are, they spin at 10,000+ rpm - ever listened to one of these beasts? Not to speak of heat, price, or the fact that nearly all other peripherals we used to plug into SCSI (scanners, CD-RW, removeable media) are today available as either ATAPI or USB devices. In fact, SCSI CD-RW drives are usually ATAPI internally with a bus adapter added to make it speak SCSI.
That leaves us with overpriced, noisy, hot server drives that offer less capacity than any IDE drive for half the price.
And again, the Amiga virus hits: People continue to believe in things (SCSI, custom chips, etc.) long after these things have been made outdated by reality... |
|
Pentium IV Prototype at 3.5 GHz : Comment 29 of 41 | ANN.lu |
Posted by Kjetil on 29-Aug-2001 22:00 GMT | SCSI ride systems, you can double the speed or the security, depending on what you wont
or even booth, count the number of harddrivers you can have on the SCSI chan and count the number of IDE drivers you can connect, this fact is important. If you own a large computer company. you can do the same with ATA if you have a ATA ride system, butt the number of harddrivers is as equaly inportent interms of GB storage,
As for boot time, my computer boot slower the any Amiga, even Amiga 500,
do not count standby, as standby is not booting.
Delays to show the Riva TNT 2 billboards,
scanning the IDE bus,
Memory count,
floppy driver seek, and so on. (optional)
delays to show the Adaptec SCSI billboards, (optional and the hot keys)
scanning the SCSI bus,
delays to show the installed option list,
(verifying DMA pool data, what is this one),
and delays on the lilo (Linux boot loader optional)
and then the OS finely load all it drivers (again).
then the OS do a new SCSI / IDE scan to se
if the driver loaded correctly and to se if the drivers exists.
This is relay about PCI plug and play v.s. Zorro autoconfigure TM.
On zorro bios and driver is the same ting, (not if you load Linux, uses it own drivers)
Amiga boot:
Run Kickstart (run aga/ecs gfx driver)
Do zorro bus, and add librarys, devices to library database (driver loaded)
Do an IDE scan (wait until ide scan is financed)
Do floppy seek if (bootpri has a grater number the IDE drivers)
(Read hard drive, Load not loaded drivers, and patches.) |
|
Pentium IV Prototype at 3.5 GHz : Comment 30 of 41 | ANN.lu |
Posted by Chip on 29-Aug-2001 22:00 GMT | In reply to Comment 28 (Solar (BAUD)): Do you know what are you speaking about?
ATA-100 -> 100 MB?????? Don't make me laugh!
Do you know the real differences betwen IDE<->SCSI? IF yes, you wouldn't speak those craps... |
|
Pentium IV Prototype at 3.5 GHz : Comment 31 of 41 | ANN.lu |
Posted by dakang on 29-Aug-2001 22:00 GMT | Easy now everyone ;-)
The ATA specs are the maximum bandwidth that the interface is capable of. Remember no IDE drive in exisitence today stresses a UDMA-66 let alone a UDMA-100 interface. However having a UDMA-100 interface/bus means that 2 very fast IDE drives on the same channel do not have to compete for bandwidth so much.
IMHO the DMA of SCSI is far better than UDMA-IDE, but for the home user rather than the business, these sort of latencies are worth it compared to the cost of SCSI. |
|
Pentium IV Prototype at 3.5 GHz : Comment 32 of 41 | ANN.lu |
Posted by Thomas Frieden on 29-Aug-2001 22:00 GMT | In reply to Comment 20 (Richie): > There are actually two companies reseaching and selling lots of X86.
> The question is "Is the ppc here to stay or is moto fixing to drop it like a hot potato.
The PPC is actually developed by two companies, Motorola and IBM....
And they both make progress in different areas. While Motorola made Altivec available,
IBM succeeded in reaching higher clock frequencies on the G4 (for a long time, it couldn't
go beyond 500 MHz).
Regards, Thomas |
|
Pentium IV Prototype at 3.5 GHz : Comment 33 of 41 | ANN.lu |
Posted by on 29-Aug-2001 22:00 GMT | In reply to Comment 13 (Graham): > The i850 chipset also supports Socket 478, btw.
Yep, I know, but at the last I checked, all motherboards I could find for the 478 were based around the i845 chipset. As of 2 weeks ago, all of Intels own motherboards that I saw were also based around the i845. Looks as though I stand corrected here now as there are now 478/i850 motherboards available.
> The tests I have seen put a 1.4GHz Athlon at about the speed of a P4 2GHz on i850 with socket 478. The P4 wins media benchmarks, Quake III and memory bandwidth, the Athlon everything else, for 1/5th the price.
Check out http://www.tech-report.com/reviews/2001q3/pentium4-2ghz/index.x?pg=1
Their conclusion is that the P4 2GHz is actually faster overall than the Athlon 1.4GHz.
> AMD will release the longer pipelined Clawhammer and Sledgehammer processors next year, which will push performance really high. Intel must be worried about that, hence their MHz push at the moment.
I don't think their over worried. When they go to Northbridge based around 0.13 micron fabrication that should allow them to bump up the old GHz a bit. Besides, lengthening a pipeline will take some of the edge out of the new AMD chips. If Intel can pull off further speed increases without further pipeline increases, then they may be in an ok position.
> Shame that the market is in a slump, eh?
Oh, such a shame :-)
<<The new P4's (socket 478) are utter shit, as they use Intels 845 chipset, so using SDR DRAM.>>
> Wrong. i850 supports both S423 and S478. i845 supports both S423 and S478.
Yep, once again I knew that both chipsets supported each of the chips, but I was basing my argument on what was available in terms of motherboards at the time. I see now that i850 m/b's are now available for the 478, so again, stand corrected.
<<Tests have shown that 2G pentiums (478) are no better whatsoever than the Athlons.>>
> Tests have shown that a 1.8GHz P4 on i845 performs like a 1.2GHz Athlon if it is lucky. P4 + SDRAMcrap.
Yep, that's what I was basing my original argument on. Looks as though the 478 is pretty successful with rambus though.
<<But in 6 months time, there'll be another revision of the Pentium which'll use DDR DRAM.>>
> You mean now, don't you? And it isn't the processor that supports RDRAM or DDRRAM, it is the chipset. VIA P4X266 supports PC2100 DDR RAM. i845DDR only supports PC1600... SiS645 supports PC2700 DDR RAM!
No, what I was referring to is that Intel were not releasing licenses to use DDR supporting chipsets. There was speculation saying that this would probably become available when Northbridge chips become available. I think it was Via who were saying that they would produce a compatible chipset without a licence from Intel and worry about it later (at least, I'm pretty sure that was correct as of two weeks ago).
> Yes, and AMD will have 1.53GHz Palomino Athlons out in the first week of October, which will be equivalent to a 2.2GHz P4. Palomino supports SSE and prefetch, hence it will have even better performance.
2.2Ghz P4 are already out. By October I reckon that we'll be seeing even higher clocked Pentiums. Interesting to see which'll be the fastest in real terms. Then we'll see Northbridge (and I've seen speeds touted as being in excess of 3GHz).
>Increase in memory bandwidth can increase framerate. Usage of SSE and SSE2 instructions can increase framerate. Increase of MHz when using the same processor can increase framerate.
I totally agree, but MHz (or GHz as the case may be) is an easy number to sell to people. In terms of performance, the proposed increase in P4 cache sizes may also go quite a way in offseting the new Athlon technology. From what I've seen so far, Intel have the edge. Time will tell if they keep it... |
|
Pentium IV Prototype at 3.5 GHz : Comment 34 of 41 | ANN.lu |
Posted by Hans-Joerg Frieden on 29-Aug-2001 22:00 GMT | In reply to Comment 28 (Solar (BAUD)): My Amiga re-boots the workbench in exactly 12 seconds. If I start my PC and Amiga at the same time, even with the enforced reboot of setpatch my Amiga is the first to be booted, and that is even if I leave Linux or Windows at the login prompt.
Concerning SCSI: ATA can only connect four hard drives. If you have a burner and CD-ROM you're down to two. SCSI can connect 15 devices, *including* scanners and other external stuff. My SCSI disks aren't any louder than the IDE disks. Only a lot faster. As you said yourself, ATA-100 is completely irrelevant since no hard drive currently is that fast (see comment 6).
The only pro point for IDE is the price.
I would have expected better arguments from you than invoking that same old "Amiga virus" topic. Just because one is an Amiga user doesn't mean you're stupid. |
|
Pentium IV Prototype at 3.5 GHz : Comment 35 of 41 | ANN.lu |
Posted by Ian Shurmer on 29-Aug-2001 22:00 GMT | In reply to Comment 28 (Solar (BAUD)): Surely the BIOS stage must be included as part of the boot time; because if
the hardware wasnt initialised within this stage then the installed OS (Windows in this case) would have to do it anyway.
Ian |
|
Pentium IV Prototype at 3.5 GHz : Comment 36 of 41 | ANN.lu |
Posted by Tony Gore on 30-Aug-2001 22:00 GMT | In reply to Comment 23 (nian): I agree partly with the fact that you need a fast graphics card like Geforce2 or Geforce3, but the CPU does play a part in your gaming experience. Just take a look at Croateam's "Serious Sam" (http://www.croateam.com). It is a fps, with single player, co-op, or deathmatch via internet or lan. This game has some really huge playfields and an enormous amount of activity on the screen. (Wish Croateam would come back to the Amiga market and port this for the new Amiga's) It's a really great fps game, but on lower MHz systems, just doesn't play as good. Even on this Ghz PC with 256MB Ram, it studders sometimes when playing online competition. So looking at what can be done with the hardware that's available, you can see why the push for even faster hardware. Although I'm sure a ported version of this game would run good on the current higher end PPC's with a good gfx card. |
|
Pentium IV Prototype at 3.5 GHz : Comment 37 of 41 | ANN.lu |
Posted by Thierry on 31-Aug-2001 22:00 GMT | In reply to Comment 34 (Hans-Joerg Frieden): Hi,
I was wondering, 12 seconds. What CPU and speed is your Amiga?
I've always wondered, does a SCSI Compact Flash card reader exist? If one does, could it be hooked up, set to ID 0, and a Compact flash card of 16 Megs could almost make an "instant on" Amiga? You could even put multiple boot roms on it, although illegal; I did email Amiga Inc. asking for this to be sold by them.
I remember reading, a long time ago that 68040's were booting too fast, the hard drive didn't spin up in time to feed it information, some kind of delay had to be implemented, but, however, a compact flash card would be ready the second the power came on. The only question is, what is the read/write transfer rate of Compact Flash cards? BTW, I think they're the best thing out there.
I would do this myself, but my budget, currently is in the gutter, so, I'll have to leave it for another day.
Amiga is STILL simply THE best!
Vancouver, B.C. Canada |
|
Pentium IV Prototype at 3.5 GHz : Comment 38 of 41 | ANN.lu |
Posted by Thierry Atheist on 31-Aug-2001 22:00 GMT | In reply to Comment 34 (Hans-Joerg Frieden): I wrote in comment 37
/*
I remember reading, a long time ago that 68040's were booting too fast, the hard drive didn't spin up in time to feed it information, some kind of delay had to be implemented, but, however, a compact flash card would be ready the second the power came on. The only question is, what is the read/write transfer rate of Compact Flash cards? BTW, I think they're the best thing out there.
*/
I know you said "reboot"; But, you still might be losing 2 to 5 seconds if some internal code is telling the CPU to "delay rebooting until the hard drive is spinning fast enough".
Amiga rules the OS universe! |
|
Pentium IV Prototype at 3.5 GHz : Comment 39 of 41 | ANN.lu |
Posted by sutro on 02-Sep-2001 22:00 GMT | It's quite annoying listening to comments about IDE superiority(!!!) to SCSI. SCSI is clearly the superior system and you can clearly see it in everyday use. My SCSI-only PII server at work actually feels a lot faster than a PIII IDE based system.
You can say "IDE is cheaper" but not "IDE is better". |
|
Pentium IV Prototype at 3.5 GHz : Comment 40 of 41 | ANN.lu |
Posted by sutro on 02-Sep-2001 22:00 GMT | In reply to Comment 28 (Solar (BAUD)): :: No. If you count off the HW initialisation, which isn´t a factor of CPU or OS - AmigaOne´s will have to run through that just the same. A bare-bones Win98SE takes less than 20 seconds to boot, at least on my machine - haven´t had the opportunity (or time to kill) to do throughout testing on this.
Time to buy a new watch ;-) Seriously, start installing some programs and you will see the "great" Wind0w$ crawling at boot-time like an 040-AGA-based A1200 trying to run Quake. It's suprising how many consider adequent (some believe it is good !!!) an OS that gets slower everytime you install (NOT run in startup) a new program...
:: SCSI drives are either none the faster than IDE drives, or if they are, they spin at 10,000+ rpm - ever listened to one of these beasts?
I am afraid you are wrong. IDE drives are quite good in squential I/O but, in everyday use, be prepared for a lot more system bogging than SCSI. Provided of course that you are an (ex ?) Amiga user that is used to deal with many tasks simultaneously. |
|
Pentium IV Prototype at 3.5 GHz : Comment 41 of 41 | ANN.lu |
Posted by Mike Pearson on 02-Sep-2001 22:00 GMT | In reply to Comment 40 (sutro): Various comments Re: to Previous comments. A Hodge-Podge.
Yes it is true that the 040's were too fast for the harddrive to spin up and now I have an '060 with a 10,000 Rpm seagate and it takes 3-4 reboots to give the harddrive time to spin up, it took 3 reboots for the IBM 7,200 rpm to spin up.
This is compounded by having OxyPatcher which requires a reboot to put the patch in the first command (?) space.
SCSI's are much better though the true advantages are lost on the Average consumer, and this includes most business. Not only is the transfer time better, but CPU usage is less because the hard disk has it's own chip, this is one of the reasons it is more expensive. If you are using other cpu intensive programs, this would be a benefit.
Also, your IDE transfer speed is only as great as ypur slowest IDE device. That is if you have an IDE ZIP drive attached, it isn't even ATAPI and hence any device attached to that bus will transfer only at the bus speed of the original IDE spec that rules the zip. That's one of the reasons some have an add-on extra IDE controller. That's why some of us buy scsi ZIPs.
My Seagate is quieter than any of my PC's and my Amiga 2500 case is never on. (I have no idea where it is anymore) Boy do they run HOT though, once it's been on for an hour, you can't pick it up, hence PC users should have extra coolers.
I also have replaced my scsi's less than my IDE drives. (I hope that's not a jinx!) |
|
Anonymous, there are 41 items in your selection |
|
- User Menu
-
- About ANN archives
- The ANN archives is powered by #AmigaZeux. It was updated daily (news last: 22-Oct-2004; comments last: 18-May-2005).
ANN.lu was created, previously owned and maintained by Christian Kemp, www.ckemp.com.
- Contribute
- Not possible at this time!
- Search ANN archives
- Advanced search
- Hosting
- ANN.lu was hosted by Dreamhost. Sign up through this link, mention "ckemp" as referrer and he will get a 10% commission on any account you purchase.
Please show your appreciation for any past, present and future work on ANN.lu by making a contribution via PayPal.
|