Slashdot Mirror
700 MHz Athlon
News.com is reporting that AMD will release a 700 MHz Athlon this week, pushing them even further ahead in the MHz race against Intel. I hope it works well, because the article also points out that AMD is still not doing well financially...
that AMD needs really badly at this point which would probably turn the company around.
1. A better relationship with Microsoft- Microsoft you say? How can he really think anyone needs to be friends with Microsoft? Well you need to take into account something Intel realized a long time ago. Microsoft will not go away, kinda like gum stuck to the bottom of a pair of Airwalks. Linux is great yada yada, but the linux geeks of the world don't have the mass buying power that PHB's have. 3D Now extensions in M$ products like office and maybe some special support in NT and 98 for it also.
2. An intelligent marketing director- AMD is missing something that Intel has and it's not money. According to Intel's commercials if you buy their processors you can actually move into the internet and cute bears will walk around your screen all because of the P3. All AMD has to do is get people to believe they can do some extraordinary stuff with the Athlon and when it comes down to buying a P3 system or an Athlon system Jonny and Suzy Nontechsavvy will buy the Athlon.
3. Some kind of relationship with any computer manufacturer that has TV commercials- People watch alot of TV so when they see a Dell or Gateway commercial with the AMD logo they'll think the processors are damn cool, like they do now with the P3. Intel makes a majority of their megabucks selling though computer manufacturers, learn the lesson AMD.
4. Really snazzy SMP- This fits in with all of the other suggestions. Make the SMP run really snappy with up to alot of processors, companies like Dell and Compaq will put them in their servers and then sell NT on them and put them in commercials. If AMD can keep prices low we might even see consumer SMP boxes for under 2000$ with any distro of linux or Windows NT or 2K from a said computer company. SMP is sorta like RAID in that it would work wonders in a consumer environment but never panned out and has for the most part been relagated to enterprise class machines. SMP just leaves so much room for expansion on a system, if you really need more power you can buy a second processor and you're set, no worrying if you can find a slightly faster chip to work in your system or if you just need awhile new system. Thats why I got a dual motherboard, when I need more speed I want to pay 200$ for a second processor, not have to upgrade to a while new motherboard AND faster processor. When I need a faster system still I can replace both processors and up the bus speed by 133mhz giving me a much longer life span out of my investment. This is good for Joe COnsumer. Slap the motherboard companies to get their SMP boards out AMD, you need it.
Ok that's all.
I'm a loner Dottie, a Rebel.
Posted by Nr9:
Because we already have nice and fast backside l2 caches for our processors, a faster/wider system bus won't give as much a performance boost as before.
A 25% improvement in bus bandwidth will only yield about 2% improvement in performance in modern processors with a 512k backside cache
So what does it mean to AMD if VIA can't deliver on the Apollo KX133 chipset? It means that you give Intel a chance to get back in the market with 700 MHz Coppermines in November. It also means that you reaffirm the general sense that AMD can't deliver it's products in a timely manner. Although waiting for the KX133 rewards consumers with lower CPU prices as the OEM market has demonstrated drastic drops in Athlon CPU prices this month, it prevents AMD from signing lucrative OEM contracts at a critical time.
In the consumer markets, people a waiting for AGP 4x, and will settle for an Apollo Pro chipset and a Pentium 3 rather than wait 2 months for VIA to release the same product for the Athlon. Personally, I don't plan on buying any more Intel products due to their shameless attempts to strong arm Taiwanese manufacturers into not making Athlon motherboard, the bogus Rambus fiasco and Intel's bribery of Gateway to stop making machines with AMD CPUs.
Essentially, AMD is making mistakes that they _Could_Have_Learned_from Intel. For instance, Intel makes its own chipsets, and charges people royalties to use them. AMD has virtually out-sourced its chipset design and fabrication to VIA who recently acquired Cyrix. Now, you've got to ask yourself, how dumb can the execs at AMD be? And you have to be pretty suspicious about how long it's taking VIA (pre-earthquake estimate for the KX133 release was November) to get out a decent motherboard for the Athlon CPUs, as the changes they'll have to make are hardly drastic. Current Athlon boards already have the Alpha 200 MHz bus, already support PC133 SDRAM, all that VIA needs to do is slap on the AGP 4x components and onboard UDMA/66 which is technology they_already_have. You have to wonder if the aquisition of Cyrix has made them a little less eager to see AMD survive, after all should AMD go bankrupt, that would leave the entire low end PC market at their disposal.
AMD's moving of their main Fab to Dresden was a smart play, as is the production of Motorola G4 chips (and indeed the partnership with Motorola whereby AMD has access to Motorola's copper CPU technology), However, why not shift the whole operation to Germany? Why put yourself at the mercy of VIA, and the Taiwanese?
To be blunt, AMD could put out a 2 GHz CPU by Christmas and still be screwed if VIA is significantly late with the KX133. I for one will not wait forever for this board, especially with all the information out now on the Transmeta CPU.
Athlon's current bus speed is 100Mhz.
It transfers data on the rise and fall of the clock cycle. Hence "200 Mhz".
Chas - The one, the only.
THANK GOD!!!
Chas - The one, the only.
THANK GOD!!!
Man, I didn't think they were in this much trouble. From the article:
"Last quarter, AMD reported a walloping $173 million in operating losses and revealed that Atiq Raza, one of its most highly regarded executives, had made a deal with Beelzebub and was recalled to the astral plane for punishment.
There was no word on how Andy Groves, CEO of Intel corp had managed to maintain his working relationship with the lord of the underworld.
Some cite his strict open door policy, while others think he may actually be the devil."
Hotnutz.com
Man, I really, really, really hope AMD doesn't go under. We really, really, really, need competition in the CPU sector. Competition is good, good, good. Intel has been gouging the public for long enough (not to say that k7 pricing so far isn't basically gouging).
I wonder why the gov't doesn't look into going after intel...what are the differences between m$ and intel that made the gov't go after m$ only? Technical reasons about the business? Or does the gov't just want to look like its getting something done by attacking the company everyone in the world knows of?
My understanding is that microwaves start at 1 Ghz and with CPU speed approaching those frequencies, does this mean that the CPU's will start emitting microwaves.
I'm assuming that I don't know my physics that well, because I guessing that microwave emitting CPU's would be a bad thing!
Then again, you could always cook the evening meal while you read /.
No, this isn't a flame...
What holds my interest in the Athalons isn't the tit-for-tat clockrate match with Intel (I mean, who really cares who owns the fastest CPU this month?). What I'm far more interested in AMD producing is the SMP versions.
To me, the biggest pain-in-the-ass right now is the P3 / Xeon product split. The slot1 architecture limits me to 2 CPUs - if I want more, I have to get Xeons, which are a huge cost uplift. Here's a good example:
A Compaq 1850R and a 6400R are very similar machines in everything but the CPU architecture (yes, there are a couple of other important advantages in the 6400, but not that big). I pay about $4000 for a 1 CPU 1850R, and $1300 for an additional CPU. For a 1 CPU 6400R, it's $8000, plus $1800/CPU.
Now, if AMD can come out with a chipset/motherboard that has 4-8 slots, AND I can stick the "commodity" (ie non-Ultra) Athalons in it, well, then, GoodBye Intel! AMD needs to realize that there is a huge opportunity here at the P3 / Xeon split, a place that they can heavily exploit. All the better, since Intel will NOT be able to remedy the situation, other than lower the cost of Xeons, something they are loath to do.
Having an single architecture that will linearly scale up through at least 8 (and preferably to 12 or so) would be a real advantage to AMD. I would much prefer to see them press ahead with putting into production large SMP-capable chipsets and such than worry about getting the absolute highest megahertz CPU out.
Just in case I'm not totally clear here, what I want is to be able to use the SAME CPU from the low-end machines to the high end ones. Now, I know I'll get better performance by buying the Athalon Ultras (much more L2 cache), but I'd at least like the option of using the base Athalons, rather than be forced to use the more expensive chips, just because the manufacture want me to (that's the reason for the Slot1 vs Slot2 division - it's a pure Intel marketing decision).
-Erik
There are always four sides to every story: your side, their side, the truth, and what really happened.
128-bit CPUs...yes, of course...if there isn't such thing as a TRUE 64-bit (through and through) processor, how can Hitachi already have an 128-bit one? I don't think I said that. My point was that subsystems are already digesting bits and pieces of information in chunks much larger pieces than 64-bits...aren't most memory controllers 128-bit and aren't Maxtrox cards chewing at 256-bits within the localized world of on-board video memory?
= -=-=-=-=-=-=-=-
I remember the transformation from 16-bit to 32-bit happened literally overnight. The 386 came out, was completely backwared compatible and all you had to do was run Windows in standard or 386 enhanced modes. Now, we are losing backwards compatibilty and it's not like these Merced chips are out tomorrow...wasn't this supposed to come out sometime closer to 2001? To me, this leap to 64-bit processing is more about changing the instruction set than changing the amount of data you can crunch. That's what worries me since it means that everything is going to be emulated and will probably run as crappy as my Win16 applications run in NTVDM.
Unless Transmeta fixes that but that is another discussion thread entirely...
Regarding cost...of course. But costs come down eventually as the technology becomes adopted. It may take longer to trickle down from servers to consumers but eventually it gets there. Alpha is not an accurate example because that wasn't just about "better". NextStep was "better". Better gets you nothing...if "better" also means "different" and "not backward compatible".
Time and time again, companies prove that abandoning backwards compatibility is the WORST thing you can do. Commodore killed themselves overnight by sinking all their money into the Amiga 4000, which could not use any of the software or hardware from any previous Amiga line. And NeXT, too, learned that lesson. Microsoft, I dare say, is the success they are simply because they are still dragging along that old backward compatibility mentality, and will even be doing so in their Millenium product.
All I'm saying is that I see 128-bit everywhere in marketting hype (look at Apple's new G4 descriptions) and 64-bit seems to be 1998's big buzzword. If Intel finally comes out with 64-bit some time towards the end of 2000, it won't be backward compatible so all the existing Pentium customers aren't going to be interested, and it won't be "cool enough" that power users will want to invest in it...so...disaster in the making?
If there was a 64-bit x86 processor, I would buy it but...my understand is that there is not nor ever will be if the company that invented x86 is finally washing their hands of it...
As far as speeds go, I dare say they can be the same rate...like I said in my original post, we now have busses running at 133MHz, the same speeds that processors were running at just a few years ago. You may scoff at the idea of higher bus speeds but ten years from now when my kids are reading SlashDot archives on 500MHz holographic displays imbedded in their eyeballs, they'll have to give a knowing smirk...
And regarding single point of failure...I'm not talking hot swapping nor am I talking live switchovers...this isn't about 99.9% uptime for a server. I'm talking about spending $560 on a processor instead of $125 each on four of them. If that processor dies...I'm down until I can either returned it to the store (if it happened within 30 days), return it to the manufacturer (if it happened within a year) or return it to the ground from whence it came. If one of the cheap-o processors dies, I can reboot and come back with a working system, albeit with only 1 or 2 processors.
It's like RAM and hard drives...is there anyone out there that has only one single 256MB DIMM or 20GB drive in their possession? Or more likely...you have every spare DIMM slot and IDE/SCSI connector filled even if you had to pull out a crusty 16MB DIMM or 2.1GB to do it?
Processors, IMHO, should be no different. I don't need the same processor speeds in a BeowOOOOOlf cluster, do I? I don't need the same processors speeds for SETI@Home, do I? If the architeture had been designed for SMP years ago, I don't see why processors could be as board independant as RAM or hard drives.
Okay...enough from me...the issue just hits a nerve for me because I really, honestly, have not felt love for my PC system since the 486 that lasted me for six years. Everything else seemed literally like junk the second I booted it. I think that since the Pentium era begun, I haven't been able to keep a system for more than two years...sad...really sad...
- JoeShmoe
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-- I wonder which will go down in history as the bigger failure: the War on Drugs or the War on Filesharing
The government won their suit against Intel, my friend. Unlike M$, Intel was smart enough to realize that they would lose and that the bad publicity would turn the public against them the way it has with M$. Consequently, they settled out of court for an undisclosed punishment.
I reflect your pompous signature back upon you.
That actually also implies that SMP is also of relatively little merit for the same situations, as SMP is also a solution to the CPU Bottleneck problem.
What has more merit is the notion of trying to let system components be increasingly independent of the CPU. Things like:
(Some may think I jest... Check out the buffering specs on a 16450 versus 16550...)
This would be useful in providing a way to make journalled filesystems (of which two have recently made beta releases for Linux, namely ext2 and Reiserfs) both faster and more reliable, as they could have the metadata journals written out initially to this "relatively reliable cache" rather than having it vulnerable in ordinary RAM.
1MB of battery-backed RAM would be enough for Stephen Tweedy's ext3 implementation; doubtless other uses could be imagined for larger quantities...
The increasing availability of video cards with 32MB and more of RAM implies that there's some work being pushed off to the video card.
Extra coprocessors on the video card can't be too horrible an idea, and is consistent with things like Monster 3D boards that provide coprocessors to the "main" video card...
On the other hand, there's one notable fallacy here, which is that there is merit to moving to a system working with 128 bit memory space sizes. This is silly from at least three perspectives:
This is why the only systems that presently support more than 4GB of RAM on Linux are UltraSPARCs,
There is apparently a way to support a 64 bit data bus through a longer connector slot, but that does not make for a 128 bit extension.
Based on the rates of growth in the past, it seems reasonable to expect that 64 bit addressing will be adequate for all but highly unusual applications for the next ten to fifteen years.
And even then, it is reasonably likely that by that time, technology may have progressed back to using some of the memory segmentation technologies of the '60s ala MULTICS to obviate the need for linear growth of adddress space.
2^63 is approximately 4 billion billion, which is a very large value. A move to 128 bits squares that again, which provides an even more atrociously large space.
By moving to 128 bit addressing, the complexity of the memory addressing circuitry increases, as do the minimum amounts of memory that are likely to get addressed for instructions. This means that programs get bigger even if they do no more than they would on a 32 bit architecture.
If you're not part of the solution, you're part of the precipitate.
Are there STILL people out there that think their processor is the bottleneck of their system? It's everything else!!!
As always, it depends on what you're doing. If you're like me and you spend all your time doing processor-intensive rendering operations that you've carefully designed to fit entirely in cache (and that 3D cards are incapable of performing) then the processor is still very much the limitation. In fact, I find that in most of what I do, the processor is the biggest bottleneck. Your milieage may vary.
Just keep bumping that clock rate up, AMD - we love you!
And, oh yeah, we just love that 200 MHz bus, too
Life's a bitch but somebody's gotta do it.
My biggest problem is the memory speed and that's still 133Mhz for $400 a stick.
I think AMD is pretty aware of this. After all they went to great effort to get a good SMP bus from Digital (now Compaq). However they don't make motherboards. They don't even want to be in the motherbord chipset bisness. In some ways both of these things are good. In other ways it is bad, mostly in reducing their time to market.
Look to Hotrail (I think that is a new compony name, was Posidon, or some other sea related name) for SMP chipsets. I heard they were trying to do 4-way, 8-way and 16-way ones. 4-way and maybe 8-way to be out first half of next year. I think. Ah, wait they say late 1999 for 4 and 8 way K7's.
As far as I know all the K7's are SMP capable. However the Ultra's are not just bigger cached non-Ultras. They also run the point to point bus faster, which means more memory bandwidth. I think they are planning on running it twice as fast as the non-Ultra K7's. Which means the extra bus traffic won't hurt as much.
The faster bus speed requires a cleaner eletrical enviroment the the lower bus speed. That is provided by the "Slot B" systems (shorter strighter traces I guess). Acording to AMD one could make a Slot A system that clean, but it isn't part of the Slot A specs AND costs more, so it is unlikely to be done on the Slot A systems.
My guess is you will see 2-way K7 systems, and 4-way K7 ultra systems, and later some 4-way K7's, but no 8-way or 16-way non Ultra systems. Of corse if there is a market for 8 and 16 way non-Ultra systems it doesn't look like AMD will stand in the way. It's not like the non-ultras were designed not to work in the SMP world, and it's not like the SMP chipset, or motherboards are made by AMD!!
Lastly when Intel first designed the Slot 1 vs. Slot 2 systems I thought that the PII Slot 1 module was too small to both hold the larger cache, and have enough extra cooling, which makes a real reason to have had Slot 2 in the first place. I assume that has gone away as SRAM tech has advanced.
First, a quick summary. SMP, or Symmetric Multi-Processor systems are so called because each processor is effectively equal to the other processors. There are a number of different architectures, either using clustering of procs sharing the same bus, or clusters sharing the same memory, or even clusters expanded out.
As a breakdown:
AMD SMP-to-be: processors in a 'cluster' share the same memory, I/O arbitrated by a complex chipset.
Intel SMP: up to four processors in a 'cluster' share the same bus, arbitrated by a complex bus protocol.
Beowulf - Each cluster has its own bus, I/O, but communication is much slower.
Now, one would think that performance would increase linearly for each CPU? So if you have two CPUs, your performance is 2X? Four, for 4X? Don't bet on it.
In every case, you have to factor in:
(1) SMP programming requires different algorithms to take advantage of multiple CPUs. Some tasks require too much communication, or are too linear to benefit much from having another processor. Remember that adage that you can't have a baby in one month by getting nine women pregnant!
(2) SMP split by 'thread' has limitations as well. Operating system complexity rises dramatically.. As far as I have read, the fined grained SMP necessary to make this work is still under development in all free operating systems. And this code is so complex and the tools to debug this sort of code are so expensive that it is likely to be quite some time. Linux is blessed that some large companies are contributing to the codebase for SMP support -- Intel among them if I recall. Other than operating system issues, this requires many of your programs be re-written to benefit (depending on what you want to do).
(3) In every case, there is a shared resource or cost. If every processor wants to access the memory, who "wins"? Likewise for the bus on the P6 SMP.
I've read numbers at one pointed that showed the most drop-off at four. I suspect that Intel, going for 'profitability', made four the maximum for many reasons based on simulations and probably electrical limitations.
As a final point, look at the Errata listings for Intel processors, now that they are public. Given that Intel has the resources to extensively test their processors, look at how many of these relate to having more than one processor communicating?
Now, consider that with your AMD SMP system, you have a chipset that has to handle complexities of multiple processors sharing resources, with less resources to figure out what could go wrong.
In this light, AMDs decision to focus on a single processor environment makes a huge amount of sense. If you get a huge boost out of parallelism, buy the cheapest Mhz/$ and Beowulf it!
Yes! Bus speed is an issue, but the solution is non-trivial. Those lines on your motherboard have electrical characteristics, based on length and frequency. Simply put, they become radio-transmitters! Even the layout rules for the 100Mhz/133Mhz buses are tight.
Inside the chip, it's a lot easier to push frequency because your lengths are *very* short. Thus, it's likely that there will be consumer-level 1Ghz CPUs next year or so. But the bus speed is still down near 100Mhz (even if you try tricks like double pumping it).
One solution to this problem is to put more on the same chip. These systems on a chip have been created less than happily before, as you lose most upgradeability, *but* you get to use whatever bus speeds you want internally. You also save a few bucks, and a few inches.
If I had to guess (and my crystal ball is down for the RH 6.1 Upgrade ;) ), I'd think that as cpus become a commodity item, that systems on a chip will be more and more prevalent. I'd expect alliances between nVidia and 3DFX with the AMDs, VIAs, and even Intel (despite their snapping up Cirrus Logic).
Remember, though, that if your code fits in cache and takes advance of new hints for caching present in the K7/P3, you might not even notice the faster bus speed.
My wish for improving computer performance would not be a faster bus, faster clock speeds - it would be software:
An excellent optimizing compiler!
(And, a Linux port of Intel's VTune would be fairly sweet too).
(I haven't checked lately, but last I looked there was nothing that would generate code that took advantage of P6 instructions, despite conditional moves that would eliminate many branches and give a noticeable speed boost). Here's a nice little advocacy experiment for those so inclined.. Compare the output of VC6.0 under NT (running console), versus latest Cygnus build of GCC with various algorithms, Sieve of E, etc.
The RAM is too slow. Also the biggest improvements aren't due to clockspeed but architecture improvements like branch prediction, speculative execution, parallelized arithmetic, faster instruction execution.
Question:
Will the AMD Athlon(TM) processor support multiprocessing?
Answer:
Yes. The AMD Athlon(TM) processor bus architecture
is designed to support scalable multiprocessing.
As the AMD Athlon processor evolves into a family
of processors, multiprocessor systems (workstations
and servers) based on forthcoming AMD Athlon
processor platforms are planned to become
available. The number of AMD Athlon processors
in a multiprocessor system is a function of
chipset implementation, and not the AMD Athlon
processor design. Forthcoming optimized chipsets
are planned to enable multiprocessor system
designs based on 2, 4, 8 or more AMD Athlon
processors.
[rant]
= -=-=-=-=-=-=-=-
Are there STILL people out there that think their processor is the bottleneck of their system?
It's everything else!!!
Remember when the 486DX2 came out? You could have a 486DX2 running internally at 66Mhz on a bus of 33MHz...or choose the 486DX running at 50MHz on a bus of 50MHz? What did the benchmarks say? The all-50-MHz system was much faster.
But, instead of spending their time improving the speed and data width of the bus, chip makers tripled the internal speed...then quadrupled... now the minimum is quintuple the speed and AMD thinks we need septupal (notice I conveniently skip the dirty one).
I fail to see why the bus systems and processors can't make the speed transition at the same time. The extra time in the CPU cycle can be spent improving data pipes between components and increasing the number of processors. Imagine if the standard system today was four, 128-bit processors running at 133MHz on a bus of 133MHz?
I say end the madness...AMD and Intel need to spend their time working on cheap, stable 128-bit SMP instead of this overwhelmingly-fast single-point-of-failure! Why 128-bit? Because if I'm going to have to throw always all my existing technology and go through another period where everything I need to run is considered "legacy" then make it worth my while and jump right to 128-bits! 64-bit is obsolete,even for game consoles. Hell, there are Windows CE portables can have 128-bit Hitachi processors (although again, nothing else in the unit is 128-bit).
I want a nice fat data pipe running at a cool, stable 200-300 megahertz. Not some processor that is ready to go critical and burn a hole through my motherboard.
[/rant]
As always, my opinion...not yours. Remember that before when you post flamebait or troll replies.
- JoeShmoe
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-- I wonder which will go down in history as the bigger failure: the War on Drugs or the War on Filesharing
C-Net didn't do a very good job with their "spot check" for Athlon boxes. AMD owns the *retail box* market, thus the first IBM/Compaq Athlons are going disproportionately to the retail stores, not the mail order shops C-Net checked (tho egghead.com had IBM Athlon 500MHz machines in stock the other day). I just got back from my local Best Buy and CompUSA. Best Buy had at least several IBM Athlon boxes in stock (couldn't see the labels on most of the boxes), including 650MHz machines (fat profit margins). CompUSA had a much smaller selection of Compaq boxes, but the salescritter I talked to said they'd already sold 3-4 of 'em, not bad for a recent arrival they're not advertising. http://www.thechipmerchant.com is advertising 500-600MHz Athlon CPUs and *four* different Athlon motherboards; CPU prices are already dropping in anticipation of this week's price cuts.
Anyhow, it looks like AMD is ramping up production on schedule, they have a *great* product, and at 1% of Intel's market cap, their stock looks like a screaming bargain. (Which is why I've dumped what little cash I have into it.) Only wildcards are how big a disruption the Taiwan quake will cause in motherboard production and if the Dresden 0.18 micron fab will ramp OK. AMD reports earnings on Wednesday. If they shoved enough CPUs out the door in time to meet the quarter's cutoff, things should get interesting, especially given Intel's recent 820/Xeon stumbles.
Yeah, I want the multiprocessor Athlon Ultras (ships next year) too. If they can come up with a reasonably inexpensive 4 CPU board... (drool). But until Dresden ramps up they don't have any reason to push multiprocessor systems anyhow.
For a single-user system, SMP does not make a whole lot of sense unless you are working with highly parallel algorithms, yes. If, however, you are deploying a multi-user machine, then SMP can give you a significant performance boost.
:-)
Say, for example, you are running a web server that does a lot of dynamic content generation. Assumming a decent webserver program, different user requests can be running on different CPUs simultaneously.
Another example would be a general-purpose multi-user workhorse, of the type commonly found in college machine rooms. You have 400 students logging in, reading mail, compiling, running, and debugging programs, running MatLab, whatever. That is where SMP really shines, as you have a lot of compute-intensive tasks happening at once.
Please note that your points are well-made and perfectly valid, but I wanted to make sure it was understood that SMP is not completely worthless, either.
dragonhawk@iname.microsoft.com
I do not like Microsoft. Remove them from my email address.
Since this article is a little less technical, I'd like to go ahead ad voice a couple thoughts. It is of my opinion that Intel is a much more friendlier company than had been in the past when it comes to the x86 market. Intel most definately has the dough to pour into R&D to stay ahead of the x86 market. It most definately has the cash to through out a chip similiar to the G4 in raw flops, but it hasn't. So Intel doesn't hold the top bang, the top bang for the buck,and most definately hasn't come out with anything incredably exciting in some time. Intel does hold the market share. Sure the their new Merced is building a lot of hype, but when it comes down to actual production, it sure isn't rushing things. Intel has one wonderful thing goinf for it,and that is stability. Now it isn't the most stable of platforms sure, but think about the peice of mind you get when you buy intel. Odd chip sets? Nah, its the VIA and the SiS that put out those "soon to have a patch" chipsets, and wird super socket 7 MBs. Cyrix failed misribly because it wasn't even win98 compatible till way to late. You think that SGI wouldn't have picked AMD hardware if it could have? You think that the G4 is really going to be the way to crunch numbers even though the support hardware is so undeveloped (as compared to that is). MHZ is great, terra-flops: go for it!, but if you can't run the apps and keep em running.. then whats the point. Now I amilling to bet that those pro admins out there.. that if you even remotely suggested to any of the big shoulders at your company to switch from anything Intel on the x86 side, the resistance would almost come like instinct. Intel is (in my opinion) not really the MS of the CPU world as it used to be, simply because its product is tried or true. I see it as a corvette vs. hummer deal.. the army doesn't ride around in vette's for a reason. Bortbox (there goes my karma for sure)
I do agree with your observations that system componets have NOT kept pace with CPU advances, and indeed, there seems to be wide concensus that the main bottlenecks these days (at least in the x86 world) are off-chip.
That said, I think you are making several confusing suggestions:
I would like to see people concentrate more on holistic system performance, but let's keep everything in perspective....
-Erik
There are always four sides to every story: your side, their side, the truth, and what really happened.
if ((post.score() != 1) && (post.score() 4)) { post.read() }
Nope, the beowulf and first post things aren't going away -- let's lighten up a bit. If you don't like stupid comments, you can filter out most of the stale ones, and that takes care of most of the "problem".
Frankly, the long winded "informative/interesting/insightful" posts make me weary eyed most of the time. In my book, flames and trolls are much more interesting, especially the grains of truth that they sometimes work from.
Besides, it's Rob's fault for not including a "bad-attitude" option in the recent moderation roundup anyway -- lots of the flames and trolls wouldn't be anonymous if there was room made for them to be disagreeable and contrary without shame.
--
--
E2 IN2 IE?
Ok, right now at home I am running a AMD K6-2 350.
.. in reality.. in a lot of applications my AMD 350 seems to be a bit pepier...
:) to upgrade?
... BUT, for the average user.. and the average applications.. it seems that the bottleneck is NOT the CPU..
128meg ram, 16meg TNT, IDE.. yada yada..
At work I am running a PIII 500 with about the same setup.
this is my questions, why am I not seeing a near double in my performance? I would guess that it was ram.. BUT.. I took ram out of one of the other machines, and I am not seeing a GREAT performance increase..
Sure, my computer at work can crunch SETI@ blocks a bit quicker, but
QUICK NOTE: I am not a big gamer or 3D freak..
THIS IS THE POINT.. has anyone actually used a computer that goes that fast? (700 Mhz).. is it even worth it for the average LinuxNerd(TM)
It seems like these superfast CPU's are GREAT if you have VERY spacific needs.. (3D Rendering (ie. Games))
For someone out there that has a lot of experience with a lot of different CPU's, which CPU configuration is the price to performance ratio?
Will I see more performance increase by purchasing a UW/SCSI-Controller / HD????
More Ram?
More CPU?
:) For some reason I feel like if I dump my current motherboard and go purchase a AMD-700 w/new ram.. etc. I will be disapointed....
...would you brag about a girlfriend with only one breast? "My girlfriend has a giant F-cup breast and your girlfriend only has two, evenly-proportioned C-cup breasts?"
= -=-=-=-=-=-=-=-
Silly I know, but my point is that consumers can be trained to think anything is better if the advertising says it is. If they create some cute little "Puffy, the SMP doggy" all the JoeShmoes are gonna run out and ask for dual-processor systems at Circuit City and Best Buy.
Tell them they need two processors so one can access the Internet and the other can play games, or better yet...tell them they need a processor for every application they want to run. That outta put consumer demand for SMP on the laps of those boys in marketting...
- JoeShmoe
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
-- I wonder which will go down in history as the bigger failure: the War on Drugs or the War on Filesharing
I could overclock my P3 to 750mhz if I stuck a 4 foot tall heat sink on it too :P
I'm a loner Dottie, a Rebel.
Traditionally AMD chips have gone into the low end consumer "multimedia" machine and into the boxes of gamers in the know / people who aren't ready to upgrade to intels proprietary chip hole.
Now this chip is expensive & probably needs a new MB to take advantage of it. This is an expensive upgrade - so who will go for it?
Graphics / Video people? Nope these people are most likely to be looking at a G4 Mac or SGI these days. Or at least an x86 with history of good FPU performance (intel)
Gamers? Nope, this is probably too expensive - and after the history of bad fpu and wierd incompatibilities / crash bugs that AMD has, I cant see gamers going for it.
Servers? I really can't see it. Intel are firmly established in the server space, and cant see that ending.
Office workstations? Nope, they will continue to be mid ranged intel cheaps supplied in bulk by Compaq / IBM. There is no need for this power.
Consumer boxes? Nope too expensive.
Sure it's a nice bit of kit, but who exactly would buy one? The big manufacturers won't use them considering AMD's history of supply problems.
stty erase ^H