Printable View
yes i realize the remarkable forward compatability that the old motherboards have had, but look ahead. Theres a slight chance of the PIII-T sticking with S370, but Northwood will need a new socket, which is essentially an updated P4
------------------
iamsostupid.
1st order Sapasion's OC Knights
Dont, be biased. It's getting old, and I will be forced kill you.
Now we agree..... Thats the question.Quote:
Originally posted by iamsostupid:
yes i realize the remarkable forward compatability that the old motherboards have had, but look ahead. Theres a slight chance of the PIII-T sticking with S370, but Northwood will need a new socket, which is essentially an updated P4
------------------
cel.600@928
yes, and the itanium will need a new interface, and so will Foster. All of AMD's K7x cores will be socket-462, while all of AMD's K8 cores will need a new socket. So there are only two separate motherboards to buy if you decide to go with AMD. Also, all of the K6's were Socket-7, and all of the Athlons were Slot-A. Some thunderbirds were Slot-A as well, so you didnt really have to get a new motherboard. Intel will get a lot more nastier on their next few platforms. Look how they almost succeeded in forcing Rambus onto the Pentium III with their i820 and i840 boards. Thanks to VIA, and their (terrible, but working) Apollo Pro 133A, we had a moderately performing, PC133/133MHz FSB platform on which to run our Coppermines. This also came in slot and socket models, so they helped out there. Intel and AMD both have their ups and downs, but Intel will have more downs when upgrading comes into the picture with their next few generations of processors.
------------------
iamsostupid.
1st order Sapasion's OC Knights
Dont, be biased. It's getting old, and I will be forced kill you.
Sorry but the new p3's will need a new m/b even though they are socket 370
------------------
General FaTs
Knight of Emperor Sapasion
Cel 600@1134 (Fop38-1)
ASUS CUSL2
256mb Kingmax (512mg of PC-150 on the way)
Creative GF2 Ultra
30 gig IBM 75xg
20 gig Segate ATA 66
SBLive! Value
Wassup SharkyForum Readers,
As I stated earlier, I did contact id Software, and yes, Arcadian is right, get the story clear, as I haven't done in the past. Here's the story on SSE optimizations from id representative, Dan Carreker himself:
"Dear Atif,
I'm not able to say for sure how much support [SSE} made it into the final
version of the game [Quake III Arena], however there was an interesting quote re: that very question [Amount of support for SSE Instructions in Q3A]...you can see it at http://firingsquad.gamers.com/hardwa...ium3/page4.asp
Sincerely,
Dan Carreker
Customer Support Representative"
The link he provided me says the following straight from John Carmack's mouth, and complies fully with what Arcadian has said,
"Most of the Katmai optimizations [for Quake 3] are in the OpenGL drivers. We may have some loops in the main code Katmai optimized, but it is a low priority. Because up to 75% of the execution time of the game is in the graphics driver, most of the burden of optimization is theirs. I know that Intel is working with ATI and Katmai on their drivers."
He went on to also say, and this is KEY,
"I believe that the driver guys are getting about a 25% total speedup with Katmai optimizations. Combined with the clock rate boost, that is a significant win."
According to Carmack, Quake III features little native SSE support, however, the graphics drivers, which are referenced to during the play of the game, have been written with native SSE, and according to Carmack, 3dNow! Support. This is what leads to the performance increase.
Late
Intel will realise 2 versions of the 2ghz p4 a current socket one and the new typeQuote:
Originally posted by marv:
With all the new processors you have listed coming to market in the near future, is there a planned chipset that will proved some path of upgrading for many generations of processors as this writer's BX chipset has? It appears the AMD's 760 may have this capablity, how about Intel?
------------------
General FaTs
Knight of Emperor Sapasion
Cel 600@1134 (Fop38-1)
ASUS CUSL2
256mb Kingmax (512mg of PC-150 on the way)
Creative GF2 Ultra
30 gig IBM 75xg
20 gig Segate ATA 66
SBLive! Value
Also dont forget alot of AMD mobos are crap and realy only ait makes a good 1 and alot of people have had to buy new mobos and they are ganna have to again with the DDR bords.
For the p4s The new socket will go all the way up to 2Ghz man that willl last evan me the next fer years also the chances of u getting a k8 and switching with there different new cpus is not likley and I bet AMD will need to have different chipsets for each new cpu I dont think they will all work at PEAK performace on the same chipset.Then again I dont know yet nor do u we will have to wait and see http://dev.www.sharkyforums.forums.relay.cool/)
------------------
p3 [email protected] stable 8.5 X 133 using a fop 38
ASUS cusl2c
Voodoo 5 5500 NOT OC
Aureal vortex super quad
Quantum 20gig LM ata66 HD
Dual p3 600s one cbo and 1 oco @ 840Mhz =1.68Ghz
1.75volts on each
MSI dual bord
Asus geforce 2 GTS deluxe
250/397
Sound blaster live value!
well, I might know I will not switch to K8, because I have no reason to run 64-bit code. As for Fats, yes, Abit are in my opinion the best motherboard manufactureres. I can't wait until (are they?) they come out with an AMD760(MP?) motherboard. In about 10 days I will buy their KT7A-RAID.
------------------
iamsostupid.
1st order Sapasion's OC Knights
Dont, be biased. It's getting old, and I will be forced kill you.
I just wanted to thank email_atif for going through the trouble of contacting ID Software, and to look up some interviews with John Carmack in order to find some concrete information for us regarding Quake III and optimized SSE instructions. It is very nice to finally get some confirmation for this. Thanks a lot, email_atif, I really appreicate your find. https://www.sharkyforums.com/images/.../2005/06/5.gif
Well, I do want to point out some things here. First, Itanium will be aimed at markets quite different from the average PC user. These companies are not looking for a direct upgrade path from their current solutions, but they are willing to change their entire systems for 64-bit processing.Quote:
Originally posted by iamsostupid:
yes, and the itanium will need a new interface, and so will Foster. All of AMD's K7x cores will be socket-462, while all of AMD's K8 cores will need a new socket. So there are only two separate motherboards to buy if you decide to go with AMD. Also, all of the K6's were Socket-7, and all of the Athlons were Slot-A. Some thunderbirds were Slot-A as well, so you didnt really have to get a new motherboard. Intel will get a lot more nastier on their next few platforms. Look how they almost succeeded in forcing Rambus onto the Pentium III with their i820 and i840 boards. Thanks to VIA, and their (terrible, but working) Apollo Pro 133A, we had a moderately performing, PC133/133MHz FSB platform on which to run our Coppermines. This also came in slot and socket models, so they helped out there. Intel and AMD both have their ups and downs, but Intel will have more downs when upgrading comes into the picture with their next few generations of processors.
Foster is different as well, because usually motherboards featuring the Foster (or Pentium 4 Xeon, presumably) will require some added features to make systems more server and workstation favorable. Features like 64-bit PCI and server management capabilities. Most customers interested in Foster will not have a PC motherboard, so the upgrade path will not be necessary.
As for actual PC products, both Intel and AMD have gone through slot and socket packages. AMD has had slot A and socket A designs, while Intel has had slot 1 and socket 370. However, while Intel has allowed upgrade paths for both slot and socket users, AMD decided to discontinue their slot processors, and move fully into socketed processors. Many people don't remember the frustration this caused, and others simply forgave AMD because their socketed processors gave larger performance with the Thunderbird core.
However, it is worth noting that AMD destroyed their upgrade path once before. Now, people are cross with Intel because they are going from socket 370, to socket 423, to socket 478, all in one year's time. Although this may mean a lot of changes to endure, I'm sure that Intel, unlike AMD, will continue to support an upgrade path for their older socket designs, just like they continue to support slot 1 to this day.
AMD may not have a beautiful upgrade path either. Correct me if I'm wrong, but I believe that the multipliers on current Athlon motherboards maxes out at 12.5x. This means that the 1.2GHz Athlons we enjoy today are the largest that can go on these particular motherboards.
Of course, 12.5x can go a long way, when put with a 133MHz DDR front side bus (in fact, all the way to 1666MHz), but current motherboards can only accept Athlons with the current 100MHz DDR front side bus, which will max out at 1.2GHz (or 1.25GHz, if there were such a product). The KT133a, which is one of the motherboards supporting the newer 133MHz DDR front side bus, requires a new motherboard to be purchased. I know that a lot of people on this forum plan on purchasing these boards... I wonder why they aren't participating in this discussion about upgradability in the motherboard?
The truth is, in order to upgrade past 1.2GHz, you will either need a BIOS maracle to allow current motherboards to get higher than 12.5x multiplier, or you will need to forego upgradability and buy a new motherboard. I think that AMD is in the same position as Intel, but I wonder which one is going to support older standards....
Unless they do the k6-cxt core trick again, and reuse lower multipliers as new ones???
Then again, 12.5X multiplier is very high IMO anyway, and can be a limiting factor in some applications. Much better to raise the FSB, or the data transfer rate as you get higher performance that way.