Why AMD Is Still In The Race 272
Steve Kerrison writes "Despite a woeful inability to provide some of its most loyal customers with stock, and a range of CPUs that, currently, loses out to Intel's Core 2 processors in both price and performance (and who would I be not to mention the diminishing AMD fanboy numbers?), AMD's still got enough tricks up its sleeve to retaliate against Intel in due course. HEXUS.net has an opinion piece on why AMD isn't up the creek. From the article: AMD has been showing off its 65nm wafers for a few months now, which means the Rev G core is on its way. Even if the DDR2 memory controller which arrived with the Rev F only had a small performance benefit, Rev G has a few more improvements than just the die shrink. The latter will enable higher clock speeds and a lower price, plus allow AMD to compete on an equal playing field to Intel, which has been manufacturing 65nm processors since the Pentium XE 955 at the end of 2005."
It seems they still have enough customers (Score:4, Interesting)
Otherwise, yes, Core 2 Duo is superior at the moment. I wonder if this will last when AMD goes to 65 nm.
Need to up the ante (Score:5, Interesting)
Even if AMD goes back to their old copy-Intel ways, the value they have brought to the average is immeasurable. Intel would still be stuck on their old single core processor, instead of making plans for 80 core chips that top out at 1 TeraFlop in 5 years. AMD pushed them to get there. AMD needs to focus on creating something far better, and they need to do so quickly
Give Them Time to React (Score:5, Interesting)
And please include a 'value' analysis in your report on 'ass mastering' because the lower range Athlon 64's are much closer to my price range [newegg.com] than the lowest priced Conroe [newegg.com]. You know, there's a vast market out there for people who just want CPUs that run a word processor and connect them to the internet. Vast.
No way. Intel made a comeback? You mean that whenever one side comes out with a newer chip, they are beating the other side? This completely blows my mind. Completely.
Look, give AMD time to react. I don't think many people have considered them out of the running even for a second. And don't forget about the AMD/IBM alliance [edn.com]. IBM's research (and that is a lot of $$$ & research) backs AMD.
I find your opinion article to be largely unecessary and fear mongering -- who said AMD was in trouble in the first place?
Re:It seems they still have enough customers (Score:4, Interesting)
True. But very frustrating as a VAR. To paraphrase Barabara Mandrell, we were AMD when AMD wasn't cool. We used to take a beating from customers who were Intel brainwashed, and now, now that AMD has begun to enter the mainstream consciousness, I can't get them.
Ingram Micro is the largest distributor, and they are almost always out of stock on nearly every AMD processor. So, I either have to buy them retail or use Intel.
I'm glad for AMD, but sad for me (sorry about the cheesy rhyme).
CPU is only 1/3rd of the equation. (Score:5, Interesting)
A. mid-range Intel Core 2 Duo, 1GB DDR2 RAM, and a decent motherboard
B. mid-range AMD Athlon X64, 1GB DDR RAM, and a decent motherboard
Odds are very good that you will save $50+ going with AMD. That may not seem like much, but if you skimp just a little across a whole system you can save $200+. If you want to go SLI then it gets a little trickier. I have had bad experiences with ATI, so I go with nVidia. There are VERY FEW Intel nVidia SLI boards (in fact, maybe like 5 at the max), so there is not much choice there. There are a lot of ATI SLI boards, though. AMD has nothing but nVidia SLI, so there is a large range of options. Also, the increased bandwidth of DDR2 vs DDR doesn't get you any performance boost at all right now; maybe it will in the future. I would have loved to go with Core 2 Duo, but I felt that AMD's platform just had more options.
Self-adjusting? (Score:5, Interesting)
Of course, that requires AMD to stay in business...
Re:Chipsets.. (Score:4, Interesting)
As for NVIDIA driver quality, I'd have to disagree. I think ATI is currently better off on the driver front. Look at vista. NVIDIA's drivers only work for 2d and business applications. I can't even run ET or WoW. Day of Defeat Source runs at 30fps with the stock driver and is unplayable with the latest beta driver. ATI users are reporting working games online. The ATI All in wonder series still has software issues, but the basic video card driver works fine. I think NVIDIA doesn't know what to do right now. Many ATI motherboard chipsets were taken off the roadmap as were new card releases. Its possible ATI will play the NVIDIA game and not sell any cards themselves anymore. Its possible ATI will be exclusive to AMD systems. Its hard to say.
My personal preference has always been for ATI video cards and intel chipsets. Intel has slow chipsets, but they are stable. For AMD systems, I always buy nforce chipsets.
I just built a new system last weekend with an Intel DP965LT motherboard, Pentium D 805 (yes its hot but i had a small budget), and an NVIDIA Geforce 7300 GS PCIe. My old system was a Dell Precision 650 dual 2.0ghz Xeon with an ATI AIW 9600xt. So far, my new system is much faster with disk io and cpu bound tasks. (expected with sata and faster processor) The video framerate is poor with the NVIDIA card. I expected to do about the same (60fps in ET and 30-45fps WoW). I did buy a budget card, but I find it interesting the latest generation can't even keep up with ATI's 9000 series. With ATI gaining AMD's fabrication facilities, this could be a final blow to NVIDIA. I bought the NVIDIA card because there are FreeBSD/MidnightBSD drivers.
On a side note, anyone looking at that intel motherboard should google it first. There are some serious bios issues intel is working on and its very picky about memory chips.
I was under the impression AMD and NVIDIA collaborated on the first nforce chipset.
Re:Chipsets.. (Score:4, Interesting)
I don't know why AMD chose ATI outside of being cheaper. Nvidia practically saved AMD with NForce, especially in the corporate sector, where most companies wouldn't touch AMD with a ten foot pole because of os stability problems the other chipsets would cause. I don't think ATI is even close to making a chipset remotely competitive to Nforce Stability wise. At least their current graphic drivers don't suck as bad as they did in 99.
I remember having my K7N420 and nothing could compare to it in stability wise during it's day. it blew my older KT7A completely away when it came to uptime, in fact it's still is used as my secondary today. It needed a capacitor replacement at one point but MSI took care of that out of warranty a few months ago. Both My Shuttle sn41g2's have been rock solid as well.
If I had to make an honest guess what is keeping the AMD fanboys away, it's the sockets. I know it's my big reason. One thing I could trust about AMD is that they would support a socket until the cows come home. Look at Socket A, or even Socket 7 for example. the only time before AMD64 that they stopped supporting a socket prematurely was when they did Slot-A. and they definitely made up that mistake with Socket A.
In the Socket A Athlon Period, Intel had socket 370, 423, 478 and LGA 755. Now with AMD64, you got 740, 939, 940, AM2, and the upcoming 1207 pin socket, with talks about yet another socket revision for AM2. in the AMD64 period, Intel was phasing out 478 and was moving towards 755, and hasn't changed since.
When I used to upgrade my Athlon systems, I would start motherboard first, then processor a few months later. you could do this with no problem outside of going PC133 to DDR, now, the next architecture could be completely pin incompatible with what you buy today, that coupled with shifting RAM technologies make it a very hard sell to go AMD outside of opteron.
Hopefully, 4x4 will change this, but time will tell.
High-end is not everything (Score:4, Interesting)
#1 To create the fastest product. It makes people talk about it and therefore a lot of people end up buying that particular product, just lower-end.
#2 Media exposure. It's simple and we all know it works, but it's also expensive.
Some of you may disagree about #1, but think about it for a second. A majority of all reviews online and offline first and foremost cover the high-end products even though only a few of us can afford it. This is why the market offers products like Crossfire, SLI, FX and Extreme.
You can't buy processors in a vacuum (Score:5, Interesting)
The last Intel processor I had was a Celeron 700, years ago. I've been an AMD man for a while now. I was considering advocating, if you will, the new Intel chips until I got motherboard sticker shock.
In then end, I'd go with an AM2 motherboard and whatever processor you can afford. You're still going to need DDR2 ram, but AM2 looks to have some staying power and it accomidates the whole gambit of processor options.
Re:Sure... (Score:2, Interesting)
Re:Chipsets.. (Score:4, Interesting)
Protip: The 9600XT was an upper midrange card in its day. The 7300GS was *never* meant to be more than a minimum-cost budget solution. ATI's current budget solutions are more or less comparable to a 9600, as well. Really, you're choosing a poor psuedo-comparison to prefer one brand over another.
The last several years, NVidia's offered a better price-to-performance ratio in the midrange of the market and been very competitive in the upper range. Compare a 6600GT to the X700 Pro that was its direct competition in terms of both performance and cost. With product offerings like that, NVidia won't be going away for a long time.
They have on every NForce chipset. AMD's already very much gone out of their way to say that that won't be changing because of the ATI buyout.
Iram (Score:3, Interesting)
http://en.wikipedia.org/wiki/IRAM [wikipedia.org]
Re:Sure... (Score:4, Interesting)
It wouldn't be so hard, theoretically, to have one proecessor load all the coee from disk into memory, and hand off execution of any code written for the other processor for it to execute. The big deal would be marking which bits of code are meant for which processor. At the ELF executable level, this is already done. So it'd be possible to have an OS and applications installed with each applicaiton compiled for the processor better suited to that application. On the downside, the optimization, installation, and adminsitration of that OS and its applications could be a logistical nightmare. On another downside, or maybe on the upside actually -- we'd have to have firmware that supported both kinds of chips and prepped the sytem for them to run side-by-side when it boots. Bye, bye PC BIOS. Hello OpenFirmware or some equivalent.
If one platform came together that always used one socket for Opteron/Athlon64 and one for Power7, and someone writing to that platform could count on those two chips both being installed when someone referred to that platform, it'd be somewhat like the hardware of the Amiga with its semi-standard bevy of specialized processors. A high-end workstation that has two dual-core or quad-core Opterons, one Power7, and one really killer graphics adapter using a 4-socket board would be sweet. Maybe one of the Opterons and two Power7 chips? A base system of at least one Opteron and at least one Power7 processor with room to grow on either front would make either option feasible. And yes, it'd run Linux. In fact, Linux would probably be the first OS that could be made to run on it.
After thinking about this, I really want one. Substitute UltraSparc anywhere you see Power7, and if I'd want that, too. The strengths of all three might be nice, but in the long run it's probably better if just two that are good at different things become a platform together. Now someone just has to figure out who can and would build it. I'm not a hardware design guru, so I can't do the nuts and bolts on it. IBM could. Sun probably could. AMD probably could given enough rescources. Surely AMD and one of the others together could get it done. I hope someone does. The secret would be to either have it be a platform spec, or to have enough boards like it shipped to kernel developers that it becomes a solid option that way.
Opteron + Power7 + ATI GPU = killer workstation, and hopefully there's enough market for that to get it made.
Intel's "Lead" - burning the candle at both ends (Score:2, Interesting)
AMD's 90nm chips are *PRETTY CLOSE* in performance/heat dissipation to Intel's 65nm chips, and they completely destroy Intel's 90nm chips in both performance and performance per watt.
With there being a physical limit to how small you can engrave transistors on silicon, Intel is just rushing to that point way faster than AMD. So whenever Intel hits a sticking point on process technology advancement, AMD will still have 1-2 generations of process technology improvement (since they're "behind" manufacturing-process wise).
Basically what this says to me is that AMD's design(s) are still far superior, and they still have a LOT of headroom if they can extract the same performance gains from die-shrinks as Intel has.
Re:Sure... (Score:3, Interesting)
We should all know by now that it isn't necessarily technical prowess that leads to market dominance. If Intel is shown to be a monopoly in this case, it would mean that AMD was somehow prevented (illegally) from competing in the marketplace.
Consider this: You and your friend Timmy are trying out for the 7th-grade basketball team. You can either compete based on your merits or you can kick Timmy in the ankle, causing a sprained ankle earlier in the day. Some less-than-ethical types (Microsoft, Standard Oil) would argue that both are reasonable competitive practices.
Others would say that while Timmy wasn't able to compete in this particular match-up based on his merits as a basketball player, it was due to "illegal" competitive practices.
Re:Sure... (Score:3, Interesting)
They've got the licensing on ZRAM, which I think is by itself enough to build a better processor.
Oddly enough, though, I really wish ARM would beat out a high-frequency chip (something in the multi-GHz range). For some reason, my Pocket PC with a measley 624MHz processor can run Playstation One games faster and more reliably than my 933 MHz Dell; I don't know how that counts as a benchmark, but I've seen similar things happen with ARMs; pound for pound, they always seem to be a faster chipset.
Of course, the entire world is entrenched in x86. Not that I think it's a shit processor, or anything. It gets the job done. It's just that ARMs could do it better, given the same R&D money.
Re:Sure... (Score:3, Interesting)
Incidentally, AMD won in the non-theoretical past because intel did sit still. The quickest road to the #2 spot, when you're in the #1 spot, is to act like you're in the #1 spot. You can't quit trying hard simply because you're ahead; other people try harder when they're not #1, and they will eat your lunch.