alan81 - I don't think I agree with your interpretation. It sounds to me like the new BIOS and tuning have taken its performance beyond his expectations!
Paul

A new bios and some better tuning from our part just got gave this Quad machine wings... So I had to retest a lot of benchmarks, it spitting out incredible numbers.

Johan De Gelas
- Senior editor at Ace's hardware

wbmw, well nobody's perfect! ;)
Paul

sgolds, Andy - seems like a little mix of whistling thru the graveyard and sour grapes, doesn't it? Accentuated due to the new delay of Prescott samples maybe?
Paul

sgolds, ixse - IIRC, the latency with the extra hop through the "main" CPU is still less than a NB setup, isn't it?
Paul

ixse - right you are, forgot about the 64FX being a 14x, not 24x. However, on the RAM, I wasn't thinking about gaming per se, but on some workstation apps, etc, that might be able to use the extra memory.
Paul

EP - Perhaps you ought to read the whole review. He stated several times that it was not only fastest in gaming, but in all the other benchmarks they threw at it.
Paul

Keith - WOW. What would that 2nd CPU and a full load of RAM bring to the table?! Interesting in the Final Words section, he kept repeating fastest system ever tested, like he was looking for better words to describe it.
Paul

Oh get off it, Jerry R, Intel has almost as many radical fans as AMD, and if it were the underdog, it'd have more. Maybe you haven't been exposed to them, but I have - they're out there. Does this mean Intel can't be taken seriously? Better sell that stock!!
Paul

CJ, I'd be surprised if anyone who matters or knows anything about them would label Ace's as a fanboy site. I think Johan goes out of his way to be fair and thorough.
Paul

EP - Ah, but P4EE and Xeon aren't quite the same thing are they? I don't see any Xeons offered that have 3.2GHz freq. AND 800MHz FSB AND 2MB L3 cache, do you? And there are certainly no A64s with dual channel memory. Apples to apples please.
Paul
http://www.intel.com/products/server/processors/server/xeon/index.htm?iid=sr+xeon&
http://www.intel.com/products/server/processors/server/xeon_mp/index.htm?iid=ipp_srvr_proc_xeon+xeon...

EP - Well, I guess that explains that, doesn't it? It's getting easier to get lost in all the acronyms nowadays! <:)

Paul

Fascinating - from that same seminar, seems nVidia is going with 3GIO for graphics rather than PCI Express, like everyone else. Have to question the wisdom of fighting the crowd on this one. Shades of Rambus come to mind. Anyone know offhand if 3GIO is superior to PCI Express for this sort of application, and is it enough superior to warrant fighting the crowd?
Paul
http://www.darkcrow.co.kr/Review/Review_Content.asp?board_idx=173

Arrival of Big Daddy is imminent, it seems. Interesting memory configuration. Really quite a bit of interesting info from this Korean reseller seminar.
Paul
http://www.darkcrow.co.kr/Review/Review_Content.asp?board_idx=173

Keith, I imagine it's related to the phrase "Point of Sale" rather than "Piece of S***". Commonly used term in retail environs. Good for a giggle nonetheless. :)
Paul

Looks like ATI's on board(no pun intended).
Paul
http://www.xbitlabs.com/news/chipsets/display/20031128185913.html

OT sgolds - mostly secular? Not around here. Just look at the name! Granted it is becoming more so as this nation unfortunately does the same, but it ain't mostly so yet! Hope you enjoy yours!
Paul

EP, how can you say we have no evidence that AMD's 130nm process is running smoothly? I think the abundance of Tbreds and Bartons is proof of that. Now if you mean their 130nm SOI process, that may be another story, although it seems to be rapidly improving. And in checking out the OCs at Bunny's SuperPi site, it seems the FX51s are clocking higher(top 4 are 3.1-3.4GHz) than the Bartons are(top is 3114MHz)! Considering the 1M cache involved there, I think things have to be smoothing out considerably!
Paul

First OEM 755 MB I've seen. ECS, of course. It's good that they're using the 964 SB, so it'll have all the features.
Paul
http://www.watch.impress.co.jp/akiba/hotline/20031122/image/755a0.html
http://www.ecsusa.com/products/755-a.html

Jerry R, Of course you don't mention the fact that they all exercised options at the same time, do you? Besides, it's time for Christmas presents, and perhaps they found houses they want to buy before interest rates go up. Looks to me like perhaps they funded the options exercise, Christmas and house(et al) purchases in one fell swoop. I'd be a little more suspicious if it were happening in March instead of November, just before Christmas, with the economy looking up now.
Paul

Petz, I'd suspect it has more to do with BIOS than anything else. Apparently they've figured out how to tweak the CPU's memory controller better than the others. Perhaps there's some board design difference as well.
Paul

Keith, I don't mean exploding demand now, I mean if 755 takes off the way I suspect it could. In that case, I'm wondering if their manufacturing can keep up. At presesnt and in the past, there have been minor shortages, but they would be major if they had tried to supply anything more than ECS/PCChips and the tiny sliver of a line that Asus has carried.
Paul

Keith, I think the question in my mind is more whether they've got their manufacturing capacity up to snuff to handle the kind of demand they could have. They've always offered their chipsets at a discount relative to the others, so I feel pretty confident they'll be able to continue.
Paul
PS. Do you think the SiS-UMC merger has gone well enough that they will be able to handle exploding demand?
http://www.digitimes.com/NewsShow/Article.asp?datePublish=2003/11/07&pages=A5&seq=24
PPS. If they get good graphics out of this:
http://www.xbitlabs.com/news/chipsets/display/20030630124230.html
they'll have a killer integrated system!

Keith - Agreed! This is the first SiS product I can recall getting excited about. Tecchannel liked it as well. Considering that SiS has some similarities to AMD's current/former situation(low market share,low relative production capacity and low relative ASPs), this may be a breakthrough event for them. Done at the right time, too, not like nVidia and nForce2 in K7's twilight months.
Paul

Impressive. SiS 755 did well in tecchannel's Socket 754 roundup. Not head and shoulders above the others, but fighting tooth and nail in general, and got a substantial number of out and out wins. Looks like doing their homework paid off. Remains to be seen what nForce3 Pro 250 will do.
Paul
http://www.tecchannel.de/hardware/1263/index.html

Golfbum, yes, I found that out; but no mention of the quote being his.
Paul

Golfbum, I couldn't find any links referencing T. J. Rodgers ever using that phrase. I did, however, find numerous references to Jerry.
Paul

Some neat multimedia from a Slovakian site, of all places. Thanks to Epinephrine on SI.
Paul
http://www.paxus.sk/amd/video01.htm

Possible thoughts on "other" AMD-Sun cooperation? From aceshardware.
Paul
http://216.87.214.213/forum?read=105054169

dougSF30, I'd be testy, too, listening to what Dan3 has been spewing out. Semi clearly is right here, and I suspect Dan3 knows it. All you have to do is take one look to see that the "functional" parts of the fabs are clearly separated with old and future administrative building sections separating them. Semi has been directly involved in and speaks from experience on what Dan3 only postulates on. It's somewhat insulting, really, the way Dan has been trying to make his losing argument. Like Otellini trying to tell Dirk Meyer how to build Athlons and Opterons. Now, if you think Semi is a liar, that's another story, but I don't think so. I frequently disagree with his positions re: AMD, but I know enough about people to usually be able to pick out the ring of truth in the midst of a din of garbage(Dan not you}.
Paul

sgolds - LOL! I think that's 4x4, though. ;)

SemiconEng - I don't know about everyone else, but I was certainly giving you the benefit of the doubt on this one (FAB construction).

BUGGI, looks like Jerry isn't the only one who knows how to use other people's money!
Paul

AMD new Dresden FAB36 confirmed. Guess that answers the expected timeframe.
Paul
http://www.amd.com/us-en/0,,3715_10023,00.html

Edit: Ooops, sorry, missed BUGGI1000's post somehow.

Nforce3 250 Pro will support 4+4 memory configuration.
Paul
http://www.theinquirer.net/?article=12766

Dan3, regardless how Intel may be "defrauding" consumers re: Centrino, wbmw is right. If AMD wants to be taken seriously in the markets they emphasized in their "upended pyramids" slide at the Lehman's conference, this kind of statement is very unhelpful. There are any number of ways to say it in a business-wise manner. Sometimes a calm quiet word is immeasurably more effective than any amount of boisterous ranting.(Just ask Clint Eastwood ;) )
Paul

OT NaS- Thanks! It's Outlook 2000, BTW. What was the problem with Word as the editor? Of course, it just happens to be the way she uses it at work, so she'd like to here, too.

OT - If any of you guys(or gals) are very knowledgable re: MS Outlook and avoiding security issues, I'd sure appreciate your advice here:
http://www.aceshardware.com/forum?read=105053612
My dear wife wants to use it on her system for work related reasons, so I need to find a way to make it safe.
TIA, Paul

LOL, CJ! eom

Ace's Apache 32 & 64bit K8 results.
Paul
http://www.aceshardware.com/
Anyway, we can present you some new benchmarks already. As we were curious what the 64 bit mode can do for real applications, we tested the quad Opteron (still 4x 844 with DDR333) with Apache 1.3.26 32 bit and 64 bit on SUSE SLES8 AMD64.

* Apache bench: 100.000 requests, 10 Concurrent requests
* Quad Opteron 844 (32 bit): 8263 requests /s, 1.21 ms per 10 requests
* Quad Opteron 844 (64 bit): 9032 requests /s, 1.11 ms per 10 requests
*
* Apache bench: 100.000 requests, 100 Concurrent requests
* Quad Opteron 844 (32 bit): 8554 requests /s, 11.69 ms per 100 requests
* Quad Opteron 844 (64 bit): 9369 requests /s, 10.67 ms per 100 requests

64 bit Apache is about 9.5% faster than it's identical (As far as I could see) configured 32 bit Apache. That is not bad at all, considering that Apache is very memory intensive.

wbmw, sounds a lot like sour grapes. This is a big tier 1 win, which y'all have been chastising AMD for not coming up with. Rumors and nebulous suggestions of future use are different than a product announcement. It's also a clear sign of improving volumes, especially when seen in light of Sun's concurrent announcement.
Pual