Sign in with
Sign up | Sign in
Your question

Let's talk about your dream comp

Last response: in CPUs
Share
Anonymous
a b à CPUs
May 19, 2001 4:45:10 PM

I want to know the craziest computer that you can think about...
(Including a server setups, like dual proc, 2mb cache..)

More about : talk dream comp

May 19, 2001 5:27:57 PM

Here is my future PC you could say:

Northwood P4 @ 2ghz
Via PX266 or i845 - Aopen ATX mobo
512mb PC2700 SDRAM
(2)60gb ata 100 'Raid 0'
ATi radeon AIW 32mb DDR
Sound Blaster Live! platium
Aopen 10/100tx Ethernet (Cable)
16x DVD | 12x10x32x cdrw | 3 1/2 floppy
Aopen LX95 w/ Aopen 300w PS
-------
19in Aopen monitor
microsoft internet keyboard pro
Microsoft scroll mouse
Canon d1230uf - scanner
Lexmark z53 - printer

!!! Leader of the Anti Nvidia Army !!!
May 19, 2001 5:35:42 PM

Hmmm...one crazy setup I could think of:

Two 6 foot veritcal racks, custom designed to hold a 760MP mobo with dual 1.5GHz Palominos, one per every 2u of rack space. A third rack between the two, holding custom-designed 2u RAID chassis. Each RAID chassis would hold three hot-swap SCA hard drives and two hot-swap controller modules, each controller module having a SCSI connection to its host for data transfer and a serial connection to deliver heartbeat/failover signals. Each RAID chassis would have each controller module connected to a system in one of the other two racks, so that every RAID chassis was connected to the corresponding 2u hosts in the other two racks. Each 2u host system would be connected to the 2u host system in the other rack via a serial cable, once again for heartbeat/failover signaling. All hosts would be wired to a single copper Gigabit switch.

When any host system (or its corresponding controller module in a RAID chassis) would fail, the corresponding host system from the other rack (the one connected to the RAID chassis's second controller module) would take over. The hosts would, of course, be clustered, running Linux 2.4.4 with the latest MOSIX patches.

Not sure what I'd use it for, but it would be fun to play with!

I'd still have to keep my <A HREF="http://www.anandtech.com/mysystemrig.html?rigid=5092" target="_new">current rig</A> around though...

Kelledin

bash-2.04$ kill -9 1
init: Just what do you think you're doing, Dave?
Related resources
May 19, 2001 6:43:03 PM

It is a tough question since nowadays roadmaps are often changed. I can only see 3 months ahead. In three months:

1.5 GHz Athlon-4 (air cool) or 2.0 GHz Athlon-4 (Kryotech cool)
512 MB CAS2 DDR
2x 60G ATA 100 Raid0 HD
64MB DDR GeForce3

After three months, who knows? In a few years, we may be debating Pentium-9 vs Athlon-9.

**Spin all you want, but we the paying consumers will have the final word**
May 19, 2001 6:57:32 PM

Ok here goes:

Athlon4 @1.6 gigs
any mobo that has AMD chipset
512mb crucial DDR SDRAM
2 IBM deskstar 75 gigs RAID 0
SBL Platinum
TDK 12x CD-RW
Pioneer DVD (slot load)
NVIDIA Geforce 3
a T1 line
19" ctr monitor(dont care much for the LCD flats)

Oh, also my dream computer will transform to Jessica Alba each night for me when I come home from work. Yeah Baby!

:tongue: <font color=green> I LOVE INTEL. It tastes like chicken </font color=green>
May 19, 2001 7:26:07 PM

2x Palomino 1.53 ghz
AMD760 board
2x 80 gig 7200 Raio 0
1 Gig PC2400 DDR Cas 2
Soundblaster Live Platinum
21' Sony Trinitron
16x Pioneer DVD slot based
52x Creative ROm
16/10/40 Plextor Burner
Wireless keyboard and mouse
Altec Lansing ATP 5 four point surround sound
LS-120 disk drive
Ati Tv wonder USB
Ati Radeon 2
T1 line
May 19, 2001 7:46:18 PM

Dual Northwood P4 @ 2GHz
Asus dual P4 motherboard
2Gigs of PC1066 RDRAM (4.26GB/s memory bandwidth)
SCSI Ultra166 harddrives in a RAID configuration
GeForce 3 Ultra Ultra ;) 
Sound Blaster Live! Platinum 5.1
DVD player, CDRW

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 19, 2001 8:57:26 PM

Raystonn im guessing your using the I850e. Via PX266 is your DDR-SDRAM chipset. Good News it doesnt use the Plague 686b southbridge.

Well i'll have to choose between i845 or PX266. If Sis, ATi, ALi make a killer chipset ill look that way too. "Remember the SiS 735 is Faster then AMD 760"
!!! Leader of the Anti Nvidia Army !!!
May 19, 2001 9:46:58 PM

I won't be using DDR SDRAM unless they come out with a dual channel configuration and fix the read/write dead wait cycle latency issues.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 19, 2001 9:52:26 PM

I will not go RDRAM. Northwood w/ i845 & PC2700 DDR will be as fast as a Northwood w/ i850e & PC1033 rdram.

!!! Leader of the Anti Nvidia Army !!!
May 19, 2001 9:55:24 PM

"Northwood w/ i845 & PC2700 DDR will be as fast as a Northwood w/ i850e & PC1033 rdram."

Your PC2700 DDR system will give you 2.7GB/s of memory bandwidth. My PC1066 RDRAM system will be getting 4.26GB/s memory bandwidth. Your DDR system will obviously be much slower than my RDRAM system.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
Anonymous
a b à CPUs
May 19, 2001 10:19:57 PM

coo

--call it what you wish, with this machine I can make mercury flow in 3 directions at once--
May 19, 2001 10:21:27 PM

Then Lets see here. If RDRAM gets all this bandwith why doesn't highend video cards use it. Well lets see here. I can have twice the RAM then you and i would have more money left then you.

!!! Leader of the Anti Nvidia Army !!!
Anonymous
a b à CPUs
May 19, 2001 10:24:33 PM

I would definitely have to go with a Xeon processor. My friend has one and they are VERY fast!!!
May 19, 2001 10:24:45 PM

but a solution whith two channles of ddr will probaly be along soon then which would give it 5.4gb/s compared to your 4.2gb/s now which has a higher rate of transsfer your rdram with to channles or the ddr with two channles
May 19, 2001 10:28:51 PM

Just like 64-bit DDR dual channel is 128bit DDR

!!! Leader of the Anti Nvidia Army !!!
Anonymous
a b à CPUs
May 19, 2001 10:44:45 PM

Well i dont like rdram. Well i dont know i think my Dual p3 will last till IA64 desktop computing.

I only steal the princess cause she is hot ;*)
May 19, 2001 10:48:38 PM

This is all I ask for. A computer that is graceful. Is that too much to ask for. For example, it shouldn't crash, hang or anything else like that. If some corupt code is released in memory, it should be able to load a reserve kernel or something into a clean portion of the memory, then save all data in all programs, close them and clean the memory.

Once computers are that reliable and are capable of running at lower power than a video recorder like 20watts, I would get a couple of systems and wire them up to everything in the house except for primary power controls and security systems. And have one remote control or voice recognition (or both) to control everything.

Also, it would make sense to still keep the hardwired controls to use as failsafe. A single switch override will also be sensible.

That is all I ask for. And of course holograms and stuff.


<font color=red>"My name is Ozymandias, King of Kings:
Look on my works, ye Mighty, and dispair!"</font color=red>
May 19, 2001 10:59:05 PM

I hope there will be dual-channel DDR. If it works, it will beat RDRAM hands down. However, I am not sure about it because of the pin count. I wonder what will be the specs for MB and chipset for 128-bit wide memory bus. In any event, DDR will still have the advantage over RDRAM in terms of memory latency, which in many cases is the key to performance rather than the max bandwidth.

**Spin all you want, but we the paying consumers will have the final word**
May 19, 2001 11:01:36 PM

well nivida and amd are working on it togther right know
Anonymous
a b à CPUs
May 19, 2001 11:58:52 PM

my dream computer hmm

my tower would be of a naked chic
prefferably from s club 7 any one of those girls

Dual Poly
hmmm

what else?

--call it what you wish, with this machine I can make mercury flow in 3 directions at once--
May 20, 2001 12:05:38 AM

hmmm... Alot of topic are we.

!!! Leader of the Anti Nvidia Army !!!
May 20, 2001 1:08:58 AM

My dream computer? We are talking Star Trek/2050/fantasy/near future (ha ha) here?

Well, you guys can all play with your desktops, towers and rackmounts.

I want an implanted/wearable/palmtop computing interface with voice recognition, gigabit roaming wireless network interface with bio-security, solid state multi-terabyte storage (storage is negotiable v.s. good comms capability, but it is always good to have a local replica of your calendar :smile: ), neural/optical interface capability with backup voice recognition capability. Of course I don't want to change the batteries more than once a month so it had better be efficient... This may be a semi-organic implant device, or a modificaion to our own neural strata, in which case power is not so much of an issue. I guess in that case coffee is the way to overclock!


-* This Space For Rent *-
email for application details
Anonymous
a b à CPUs
May 20, 2001 1:43:57 AM

Dream Comp? Well, I just started ordering the parts for a comp that's pretty close to everything I wanted.

Athlon 1.333
MSI K7-Master-S
512 MB PC2100 2.5 Crucial/Micron DDR-RAM
Antec SX1240 full-tower
Promise Fasttrack ATA 100 RAID Controller
2x IBM DTLA 307030 hard-drives
Plextor Plex. 16x10x40x
Creative PC-DVD 12x with decoder card
Sound Blaster Live! Platinum
Elsa Gladiac 920 (Geforce3)
Cambridge Soundworks DTT3500 digital 6pc speaker set
Anonymous
a b à CPUs
May 20, 2001 2:32:54 AM

First my Dream PC:

Dual Northwood's 2GHz (At least)
Asus P4X266 Server version mobo (It has dual channel DDR)
Dual Channel 1GB ECC PC2700 (Maybe PC2100 I'll explain below) CAS2 DDR
3-way 180GB SCSI3 RAID 0 Array
GeForce3 Ultra
Sound Blaster Live! Platinum
Fastest DVD and CD-RW with a Hardware Decoder for DVD
Best speakers

Now. I just will make a few comments. There is no doubt in my mind that DDR in a single channel format will never compete with PC800/PC1066 RDRAM on Northwood or Williamette. Once you add Dual Channel DDR it gets more interesting. Dual Channel PC2700 (With 5.4GB/ps) should kick Dual PC1066 RDRAM's (4.2GB/ps) ass with more Bandwidth and lower latency. The only thing that I will add is that it is possible that Dual PC2100 DDR could perform slightly better on Northwood than Dual PC2700. Why? With PC2700 DDR, the Memory bus is operating at 166.5MHz (333MHz DDR), while Northwood's FSB is operating at 133MHz (533MHz QDR). In other words, the Memory Bus and Front Side Bus aren't operating in sync, and time and time again has proven that it is better for these buses should run in sync (Providing better latnecy cause in the case of PC2700 the 166.5MHz Signal coming from the memory has to be converted to the 133MHz signal that the FSB transmitts) So there is a slight chance that PC2100 could perform better because it would be in sync with Northwood's FSB, at least that's my theory. Only benchys could prove my theory. Does that make sense?
May 20, 2001 2:43:25 AM

"In any event, DDR will still have the advantage over RDRAM in terms of memory latency"

DDR SDRAM only has less latency than RDRAM when you're doing consecutive reads or consecutive writes. The second you start mixing reads and writes together it incurs some particularly nasty dead wait cycles switching between reading and writing, and vice versa. RDRAM does not have these dead wait cycles. Not many applications do a whole lot of consecutive reads or consecutive writes all in one batch. Of these applications, memory latency benchmarking applications are the most visible. The current suite of synthetic benchmarks completely avoids this problem with SDRAM by partioning tests into read tests and write tests. These results do not show real world application performance because they skip some major latency problems by the design of the test.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 20, 2001 2:52:09 AM

"a solution whith two channles of ddr will probaly be along soon"

I disagree. There are some technical problems in implementing a dual channel DDR SDRAM solution. One of the things that made dual channel possible for RDRAM was that the number of circuit pathways required for each channel is less than half that required for SDRAM. Thus, you can have a dual channel RDRAM solution and still have less pathways than you would with a single channel SDRAM solution. Dual channel SDRAM will be a very difficult design for motherboard manufacturers and would definately increase the price of such motherboards a great deal, if it's even feasible to do.

Dual channel DDR SDRAM hasn't been presented as something that's being worked on by any motherboard manufacturers or chipset designers thusfar. Thus, when someone does decide to finally look at it, it will be at least 1.5 years before the product is released. One other thing to note: If they ever make a dual channel DDR SDRAM chipset, it will require you install 2 DDR modules at a time, just the same as current RDRAM chipsets. One module is required for each channel.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 20, 2001 3:13:08 AM

Dual AMD Athlon4 1.53GHz's @ 2.5GHz each (With some supercooling $40,000 system)
Sweetest dual motherboard (not sure what it would be, but high FSB, 4GB memory, AGP Pro, etc)
4GB QDR PC5600 CL2 RAM
16x IBM 80GB 60GXP Hard drives (RAID 0+1)
GeForce3 128MB DDR
SB LIVE!
43" Plasma monitor (about $56,000)


---------
I am the first and only one with a 16MB GeForce2 GTS graphics card! :smile:
May 20, 2001 5:30:59 AM

Plasma monitors only run about $15,000 for a 50 inch one. Viewsonic makes a really nice one for that price.

"In three words I can sum up everything I've learned in life: It goes on." -Robert Frost
May 20, 2001 5:55:23 AM

nah - trust me they are not that last word in PC screens.

You'll do a lot better with a good large crt or custom built lcd wall.

The plasma panels will only do 1280 resolution, which is okay, but considering the real estate could be much better.

I have an earlier 42" Fujitsu plasma panel and that's fine for browsing and email in the living room, but no good for gaming. It looks fantastic for DVDs though (real DVD, not PC DVD.)

-* This Space For Rent *-
email for application details
May 20, 2001 5:56:06 AM

Not sure what you are talking about. Could you cite some references? So far all the articles I read (e.g. tomshardware) state that RDRAM is good for reading a large chunk of data, but the initial access time is long. What you said appears to be just the opposite. I have not read anything supporting your comments. Please give us some independent sources (i.e. not related to Rambus or Intel marketing) so we can check out on the web.

**Spin all you want, but we the paying consumers will have the final word**
May 20, 2001 7:44:18 AM

I would like a 4way 64bit SledgeHammer with all the bells & whistles.

Thats four processors to seemlessly handle:
- system tasks
- voice recognition
- artificial intelligence
- video/audio output

I think I'll call it 'HAL'

<font color=blue> The Revolution starts here... as soon as I finish my coffee </font color=blue> :eek: 
Anonymous
a b à CPUs
May 20, 2001 8:29:36 AM

Here's my future, i mean real distant future, computer.
It'll be a laptop, not a desktop...

AMD Athlon X @ 1.0 TERA-FLOP (1,000,000 mhz)
ASUS MOBO using VIA 747 B-52 Stealth chip
IBM 500 GigaBytes Hard Drive @ 500,000 rpm
50x40x100x Pioneer DVD-RW
ASUS Mobile GeForce7 w/1 gig RDRAM @ 100.0 ghz
Crucial 10.5 GIG RD-DD-XD-ZD-RAM @ 100.0 ghz
Mobile Sound Blaster Super Live! Triple Platinum 12.1 Surround Sound
Built-In 1 ,000 total watt 12 piece speaker (500 watt subwoofer included on-board too!)
19" 3-D Super Flat Ultra XGA^2 Touch-Screen, Smell-Screen, and Feel-Screen with 3200x2400 resolution @ 1,000hz refresh rate and 0.01 dpi

All this enclosed in a 5 mm Titanium case with gold trimmings =)
Anonymous
a b à CPUs
May 20, 2001 8:47:15 AM

Wow, notice how everybody put Soundblaster live! as their dream soundcard? Is it really that good?
May 20, 2001 9:00:58 AM

This latency issue is an excellent example of why, when guessing the correct answer on a multiple choice test, the percentage play is to pick the longest answer. The correct answer usually involves more nuances and hedges than the wrong ones do. Simple answers are often given about RDRAM latency versus SDRAM latency, but they are misleading at best. For example, Samsung (1) gives nice simple numbers showing RDRAM to have lower latency than DDR or SDRAM. Hyundai also has a simple answer, but with the opposite result. In fact, latency depends on a number of factors, most importantly system load.

RDRAM has a variety of power saving modes (Intel will introduce a laptop using RDRAM soon -- not for performance reasons, but to save power). When a page of RDRAM memory has not been addressed for a long enough time, it drops to a very low stage of alertness. The high latency Hyundai proclaims for RDRAM comes from this effect. (Incidentally, although the Intel P4 motherboard does not offer his option, some motherboards by third parties will probably offer the option of keeping the RDRAM always in a high state of alertness).

Latency is not independent of load, however. The higher the load, that is, the more frequent requests to memory are, the more likely conflicts are likely to occur between address lines. Part of the higher cost of RDRAM stems from the larger number of banks used, reducing addressing conflicts. Moreover, when the CPU changes from reading to writing, SDRAM (including DDR) has several dead cycles. A read followed by a write takes seven cycles: three to implement an aborted burst read sequence, one normal turnaround cycle to clear the read data off the bus, then three cycles for the write (2). RDRAM can change between reading and writing in a single cycle. This transition creates "dead time" in SDRAM that is not present in RDRAM.

Thus under extremely light load conditions, RDRAM has a modestly higher latency (although this can be altered at the expense of increasing power consumption). Under heavy load conditions, the latency of SDRAM deteoriates rapidly. RDRAM holds up quite gracefully under heavy usage. Of course, by definition, under light usage, few access to memory are being made, hence latency (or memory performance in general) is unimportant. Under a heavy load, where memory performance is crucial to CPU performance, RDRAM has far lower latency than SDRAM.

The high latency of DDR under load, along with the clock cycles lost every time a switch is made between writing and reading or vice versa (3), greatly restricts the effectiveness of the memory. It is typically estimated that RDRAM in an ideal system could exploit about 90% of the nominal peak bandwidth, while DDR (and other forms of SDRAM) are restricted to about 60% of the nominal peak bandwidth. Thus, for a 32 bit RDRAM data path, effective memory utilization is about twice as good as 266 MHz DDR, instead of just the 50% one would estimate from the nominal bandwidths. Now that real world memory bandwidth scores are available for both the P4/RDRAM and 266 MHz DDR/760 system, results from Sis Soft Sandra and other results amply confirm this RDRAM superiority. In fact, in actual systems, running a battery of real world programs, the 32 bith path RDRAM has up to 3 times the effective bandwidth that 266 MHz DDR on a 266 MHz bus does.

One other point about latency: it is actually the overall system latency that matters. For example, it reduces latency to synchronize the FSB and the memory, since the FSB is always ready to accept data just as the memory is available to hand it over. The i820, for example, suffered from a system latency because of the mismatch between the FSB and memory speeds that was not intrinsic to either theFSB alone or the memory alone. (This is why the i820 beat the non-synchronized products it was first compared to, but lost to Tom Pabst's overclocked BX440, and later to the i815. Both of the latter are sychronized.) Thus when Tom Pabst mentions that the 400 MHz FSB of the i850 is in "perfect harmony" with the 800 MHz RDRAM (which is actually 400 MHz DDR -- all RDRAM is also DDR) he has a point. Any 266 MHz DDR system coupled to the 400 MHz FSB of the P4 would have several major problems. Besides the lower bandwidth and higher pincount, the non-commensurate speeds would introduce an additional system latency into the DDR P4 product. The above exposition is likely too dry for many readers, so let me conclude with a sports analogy. Some players are very good at racking up statistics during "junk time" play, when nothing is on the line. When the pressure mounts, however, these players may wilt. RDRAM is a pressure player: the more heavily memory is stressed, the better it looks compared to the alternatives.

DDR is the king of junk time. Its latency looks good, so long as memory accesses are rare. Under pressure, DDR, like other forms of SDRAM, wilts rapidly.


Links cited:

(1) Samsung's simple latency numbers (pro Rambus). http://www.usa.samsungsemi.com/new/rdram-whitepaper.htm

(2) Number of cycles for transition from read to write
http://www.idt.com/news/1996/Oct96/zbt-arch.html

(3) SDRAM and DDR lose clock cycles in every transition from read to write.
http://www.toshiba.com/taec/components/Generic/WP_memor...


-Raystonn


= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 20, 2001 2:57:27 PM

Last time I looked the 37" was $27,000, and the larger sizes were even more. Where did you see that pricE?

---------
I am the first and only one with a 16MB GeForce2 GTS graphics card! :smile:
May 20, 2001 8:05:58 PM

My computer would run at the speed of thought and it would have a holodeck for computer games...

<font color=red>Did you ever wonder WHY aliens only abduct idiots?</font color=red>
May 20, 2001 9:12:17 PM

Man that makes me want to save up $20k and blow it on a computer....

---------
I am the first and only one with a 16MB GeForce2 GTS graphics card! :smile:
May 21, 2001 2:03:46 AM

Raystonn -- Thanks for the links. I'll take a look.

As for the rest of your reply, somehow I have the feeling something does not add up. I mean if RDRAM is actually better for heavier load and so on, then why P4+RDRAM fare consistently poorly in benchmark tests. A lot of those benchmark tests consist of running real-world application programs in pretty heavy load. In some cases, P4 with 50% higher clock speed and dual-channel RDRAM cannot keep up with PIII + PC133 SDRAM (or PIII with single channel RDRAM for that matter), not to mention Athlon with DDR. See e.g. http://www6.tomshardware.com/cpu/00q4/001120/p4-19.html.... I know you are pushing SSE2 optimization, but we are talking about the RAW power of the hardware here. We are talking about dual channel RDRAM vs single channel (PC133 or RDRAM), and we are talking about 50% higher clock speed. So what gives?? Is P4 holding back RDRAM? Maybe the best combination on paper for now should be Athlon + RDRAM??

**Spin all you want, but we the paying consumers will have the final word**
May 21, 2001 2:34:19 AM

Older applications are optimized for CPUs with a smaller pipeline, such as the P3 and Athlon. That is the cause for the Athlon performing better on some of these applications. An Athlon using RDRAM would outperform an Athlon using DDR SDRAM. A P4 with RDRAM would outperform an Athlon with RDRAM or DDR SDRAM in newer software.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 21, 2001 3:44:26 AM

Correction 1: not older applications, but CURRENT applications. That means almost ALL the software we have in our computers. Correction 2: longer pipeline is inherently inefficient. So even if software is recoded, there is no guarantee it will run faster in a longer pipeline machine vs a shorter pipeline machine for the same clock speed. SSE2 is nice, but other companies can adopt it as well. In fact, I think AMD may do that in near future.

Don't want to sound like a broken record, but this is the biggest gripe I have against your company (Intel) and P4. I know you are a software engineer, but you are not going to recode all those software out there, and you are certainly not going to give me free replacement of the software I (or anyone) use right now, which as you said are not too compatible with P4. While you are talking about future, why should I suffer now, getting lower performance with higher price. If and when everything (software and hardware) falls into place, then I'll be happy to go back to Intel. It may take a long while based on the fact even now SSE1 and MMX are not that popular, years after they become available.

Although you didn't say out loud, it is nice to see you sort of imply Athlon with RDRAM will trounce P4 with RDRAM for current software, even if it is only hypothetical. Hope you won't get into any trouble. Just kidding :)  Cheers.

**Spin all you want, but we the paying consumers will have the final word**
May 21, 2001 4:14:53 AM

"Correction 1: not older applications, but CURRENT applications"

Not really, older. I never use any software that's more than a year old. All of my 'current' software was compiled recently. For such older software, an Athlon might perform a bit better than a similarly clocked Pentium 4. However, the Pentium 4 is clocked considerably higher than current Athlons. Thus, the highest clocked Pentium 4 outperforms the highest clocked Athlon today even on this older software.

"there is no guarantee it will run faster in a longer pipeline machine vs a shorter pipeline machine for the same clock speed."

One cannot really compare two different CPUs on the basis of clocks alone. What really matters is instructions per second; not instructions per clock alone and not clocks per second alone. Multiply these two together and you get a formula for performance. The Pentium 4's higher clockspeed more than makes up for the reduction in instructions per clock.

"Athlon with RDRAM will trounce P4 with RDRAM for current software"

For older software, that's a possibility on similarly clocked CPUs. Unfortunately for AMD enthusiasts, there are no plans to use RDRAM with the Athlon. AMD is choosing to support the more popular DDR SDRAM. Popularity is not a key ingredient in performance. If it was, AOL would be the best ISP in the world. ;) 

Again, clock for clock comparisons are not really valid. I would say the most valid comparisons are either:

1) The best performance you can buy in one CPU (1.7GHz P4) or
2) The best performance (insructions per second) per dollar spent (Duron, somewhere near the 800MHz range.)

Option number 1 is going to be more expensive than the other offerings. You always pay a premium for the best, no matter how small the edge.

The Duron currently takes option number 2 with the best dollar per performance rating.

The Athlon doesn't really win either of these because A) the fastest Athlon currently available (1.33GHz) performs slightly worse than the 1.7GHz Pentium 4 overall, and B) the Duron beats it out in performance supplied per dollar spent.

-Raystonn

= The views stated herein are my personal views, and not necessarily the views of my employer. =
May 21, 2001 4:16:54 AM

I hope the rest of that kit adds up to $4,500 because the screen only retails for $12,000. Somehow I think I can build a better PC for $4500...

-* This Space For Rent *-
email for application details
May 21, 2001 5:48:04 AM

trying to be realistic now...
a dual athlon motherboard
a 2x1.5Ghz Athlon 4 with liquid nitrogen cooling (who knows where that will get u... maybe even up to 2x25.Ghz)
3 gig DDR2700
as many seagate 15,000rpm drives as one can fit into a raid array
win 2000 pro
geforce 3 with extra fans on the memory chips O/C of course
16x DVD drive
20x10x40 cd burner
jaz drive
removabe hard drive mount for those pesky 75gig 75GXP's
optical mouse
force feedback joystick, steering wheels,pedal
T3 ethernet card AND internet connection
soundblaster platinum 5.1
hooked up to a 2000 dollar digital surround system
and a 29" monitor or two
all to play the best game of all time.

S O L I T A R E !



"Don't be too proud of this technological terror you have created"~Darth Vader, Star wars
May 21, 2001 7:01:42 PM

You didn't even spell Solitaire right!!!!!! :frown:

---------
I am the first and only one with a 16MB GeForce2 GTS graphics card! :smile:
May 21, 2001 9:46:49 PM

Athlon @ 2.0ghz with kyrotech cooling
1gb of memory
2 18 gb segate cheetas at 15,000rpm w/raid
Kenwood 72x true-x
16x12x40 cd-rw
SB live plat.
3com 905 b or c nic just like mine
ATI Radeon 2
HP Deskjet 1220cxi Its a professional photo quality printer that does 13x19 prints
Nikon negative scanner Super cool scan 8000d

19" flat panel monitor
Logitech cordless optical mouse
Altec Lanscing Atp-3 speakers

BAN Tbirdinside, and AMDmeltdown
May 21, 2001 10:37:25 PM

SGI Octane 2
- Dual MIPS 12000A @ 400Mhz w/2mb L2 cache
- 2gb DRAM
- 18gb 10,000rpm fast/wide ultra SCSI
- 84gb (RAID 5) gigabit Network Attached Storage
- XIO Vpro V12 card w/128mb video memory @ 1920x1200x48bit (R12,G12,B12,A12) @ 72Hz on a 24" Trinitron
- XIO 2 port gigabit network card
- integrated 24bit 8 channel digital audio

I'm surprised at the lack of interest in SCSI.
May 21, 2001 10:46:53 PM

No offense but why the hell would you want a kenwood 72x trux and not a DVD drive? Also flat panel monitors = crap

---------
I am the first and only one with a 16MB GeForce2 GTS graphics card! :smile:
May 22, 2001 1:55:57 AM

take a look at this part of an atricale i copied


New Architecture Enables 6.4 GBytes/second Memory Bandwidth

The Grand Champion HE's 6.4 GBytes/second memory bandwidth represents a factor of four improvement over the 1.6 GBytes/second bandwidth delivered by ServerWorks' previous high-end ServerSet(TM) III HE. To achieve this dramatic gain, ServerWorks deploys banks of next-generation DDR memory operating in a four-way interleaved configuration. The design accommodates up to sixteen registered PC1600 DIMMs with capacities of 128 Megabytes through two Gigabytes each, for a system maximum of 32 Gigabytes, four times more than the earlier ServerSet III HE design. The design includes a sophisticated 128-bit ECC algorithm that enhances system reliability by correcting quad-bit errors and detecting eight-bit errors. The ECC algorithm also supports ServerWorks' Enterprise Chipkill(TM) Technology that maintains system data integrity, even in the event of the failure of an entire DRAM chip within the memory array. The design also supports a memory-scrubbing feature that detects errors before they can impact system operation; memory mirroring for extra robust operation; and hot plug memory that allows memory upgrades while the system continues in operation. The Grand Champion HE platform's combination of high memory bandwidth and high Reliability/Availability/Serviceability (RAS) allows Intel based servers to be used in applications previously served only by more expensive systems based on proprietary architectures.

Grand Champion HE Supports PCI-X, for 5 GBytes/second I/O Throughput

The Grand Champion HE's 5 GBytes/second I/O bandwidth delivers a factor of five improvement over the one GByte/second capacity of the earlier ServerSet III HE, and a factor of thirty-seven improvement over the bandwidth offered by desktop solutions that support only one 32-bit PCI bus. Because not all systems require as much I/O capacity as the Grand Champion HE delivers in its maximum configuration, ServerWorks chose a modular scheme that allows system designers to add 64-bit/100 MHz PCI-X channels in groups of two. Each dual channel PCI-X controller has its own dedicated 1.6 GBytes/second full duplex connection to the North Bridge to accommodate the peak traffic generated by the PCI-X buses it controls. These PCI-X controllers support hot plug PCI operation for maximum operational flexibility. The Grand Champion's North Bridge handles all cache coherency issues, and allows devices attached to any PCI-X or PCI bus to initiate peer-to-peer transactions with any other device on any other bus in the system.

``The launch of our new Grand Champion technology marks a number of industry firsts for ServerWorks,'' commented Raju Vegesna, President and CEO of ServerWorks. ``Our platforms are the first to support PCI-X, an important new high-performance input/output standard for server and workstation platforms, and first to deliver memory bandwidth of 6.4 GBytes/second, which is four times the memory bandwidth of our previous high-end solutions. Servers are all about bandwidth, and systems that use our new design will offer far more bandwidth than anything else in their price class.''

here is the link if you want to look at it
http://biz.yahoo.com/bw/010227/0271.html
!