Date: Mon, 3 Jul 2000 07:47:51 -0500 From: "Josh Paetzel" <jpaetzel@hutchtel.net> To: "Generic Player" <generic@unitedtamers.com>, <questions@freebsd.org> Subject: RE: amd k6-2 550 vs p2 300 Message-ID: <NEBBIJCLELPGBFNNJOFHEELECDAA.jpaetzel@hutchtel.net> In-Reply-To: <3960370A.96F8E5CC@unitedtamers.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> > > ummm...couple of things. First, I forgot to mention that I am > running my > > p2-350 at 425 mghz. That would have the cache running at 1/2 > of chip speed, > > or 212 mghz. The K6-2 runs it's cache at 100 mghz. Given that > cache hits > > are about 98-99% with modern processors I would say that is the major > > advantage slot 1/slot a/socket 370 solutions have over socket 7 systems. > > > > Second....my major performance benchmark is 3d games....most of > my Freebsd > > stuff runs on old 486s and pentium 100-166 type stuff. The > windows machine > > gets the AGP video and P2/P3 type stuff...(games are what > computers are for, > > after all. :) > > > > Third...I don't have alot of experience with the K6-2s other than seeing > > some guys run them as gaming machines. > > > > What do you mean when you say, "windows seemingly intentionally > dogs with an > > AMD chip."???? Are you trying to say that an intel socket 7 > processor runs > > windows better than an AMD one? I don't think intel has a socket 7 > > processor that is the equivalent of the K6-2. Are you trying > to say that > > AMD K6-2s run slower than P2s under windows? If that is the > case, I think > > you are seeing the results of a real world benchmark. > > > > Josh > > > > What I mean is benchmark a k-6 II and a P2 on windows, then benchmark > them on a real OS. Windows runs like crap on K-6 chips, far slower than > it should, where as on FreeBSD a k-6 II and a p2 perform roughly the > same. And where did you get the idea that slot chips were better? What I said was slot one/slot a processors have a better (faster) architecture than socket 7 processors do: "I would say that is the major advantage slot 1/slot a/socket 370 solutions have over socket 7 systems." (here's the exact quote) > > Its > the other way around, that's why everyone is going back to sockets. On > a socket CPU the cache is on die and runs at full speed. That is true of socket 370, but not of socket 7. Cache runs at motherboard speed on socket 7. (which is 100mghz for the k6-2) A slot 1 P2-350 would run the cache at 175mghz. > With a slot packacge the cache chips are sperate on the card and run slower. Yes, slot 1 cache is slower than socket 370, but it is still faster than socket 7 cache. > Are you thinking about the extra cache on the motherboard perhaps? No. Notice that I included socket 370 on my list of things that were better than socket 7. That is one of the sockets that, "everyone is going back to." isn't it? MS has > long been in bed with Intel, AMD products always perform better on > non-MS software. It still shouldn't be that much slower though. But > keep in mind you are running a benchmark written for intel CPUs, which > might use SSE, and is checking FPU performance. A real world benchmark > is actually using the chip for what you want it for, in this case fire > up UT and see what kind of frame rates you get. > As I indicated, games are the only benchmark I am interested in. I should add that UT is the primary game that I play. For the things that I do with FBSD I don't see alot of difference between a K6-2 and a P2, either. But then, I don't see much difference between a classic pentium 100 and a P3-600, either. Most of the things I do with FBSD put very little load on the CPU. Mostly I see the disk subsystem and the memory subsystem being worked. For instance, I have the old www.stomped.com web server sitting here, and its a K6-233. (stuffed with RAM, though) I didn't get the disks, but I bet they weren't 5400 rpm IDEs. ;) Windows IS a resource hog, and it does use a lot more CPU time than FBSD. Maybe that is why you notice a big difference in performance between OSs. I have little love for M$crosoft, but I find it hard to believe that they deliberately mangle the OS to run slower on a specific chip. Josh To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?NEBBIJCLELPGBFNNJOFHEELECDAA.jpaetzel>