Originally posted by: Askheart
While the hardware reviews put the KT333 against i850, was DDR333 ever supported by JEDEC?
DDR333 has been a JEDEC ratified standard for quite some time now, and their reportedly nearing completion for a DDR400 standard. Initial DDR333 chipsets did support it prior to jEDEC standards being set in place for it however.
Just to clarify things! There is no such thing as a hyper threaded P4. Yeah the Xeon has it and even with hyper threading it still lost to the AthlonMP. Hyperthreading also doesnt help performance of single processor apps much.
Okay I admit the p4 is doing a lot better than it used to in the benches especially in SYSmark 2002 for some odd reason. I still think that Barton armed with the 333FSB and 512 cache will return the AthlonXP to dominance. I think that its october launch isnt that optimistic since if you studied the Thouroughbred core you would see that it was designed for the purpose of adding more cache.
Now everyone who disagrees with me. Get this through your thick heads. My point AXP+DDR vs. P4+RDRAM almost every benchmark comparing the two uses this combo. There are many DDR chipsets for the Athlon as well as many for the P4. We all know that there is 850 and 850E are the only two chipsets out right this very second for the p4. RDRAM offers a lot of memory bandwidth that the p4 can take advantage of unlike the Athlon can. DDR looks to be the offical standard for the next couple years and the overwelming majority of all PC's sold use DDR. (NOW here is the 2nd part of my argument) Someone who knows nothing about PC's wants to buy a brand new PC. He sees a PC with an Intel 845g and a 2.26ghz P4. A not uncommon setup for a OEM PC sold. Now it uses the Intel onboard graphics and DDR 266. He sees another PC an AthlonXP 2200+ with a Nvidia Nforce chipset DDR 266 which is not an uncommon setup. He doesnt know which to buy! Lets say they are equally priced as well if not the Athlon being cheaper. He looks online to see which computer is faster! He stumbles along on these hardware sites. HE sees the P4 crushing the AthlonXP in areas of memory bandwidth and games. He assumes the p4 is the best buy. Little does he know but his p4 doesnt use RDRAM which gives it the advantage in the various tests. Not only that but his Intel "Extreme Graphics" are eating his memory bandwidth as well just plain sucking. The benches he saw all used RDRAM 1066 which is not used in any p4 systems sold by major manufacturers. He should have picked the Nforce setup we all know that. This guy isn't a hardware entusiast he doesnt even know what a chipset is! <~~~~This guy is your average pc buyer and he makes up 97 percent of the market. Our small niche market doesnt really make a dent into te billions of dollars that the major PC companies make. We know what is good and he doesnt. HE sees his processor and another processor compared and thinks that is the only difference. HE doesnt know that your RAM or motherboard chipset can affect your processors performance.
p4 system performance right now is unequal with PC1066 and I for one would like to see it compared to the AthlonXP. Though I would also like to see what the overwhelming majority of PC buyers are getting their hands on. The value OEM pc for the most part sold is a much better deal with an Athlon setup. Thanks to the Nforce chipset and the KT333/266A. Why can't benches show p4+ddr, p4+PC1066, and AXP+ddr? If I said lets compare an Athlon with SDRAM vs a P4 with DDR most if not all people woluld say that was unfair. Well comparing an Athlon with DDR vs a P4 with PC1066 is not that much different.
Okay I admit the p4 is doing a lot better than it used to in the benches especially in SYSmark 2002 for some odd reason. I still think that Barton armed with the 333FSB and 512 cache will return the AthlonXP to dominance. I think that its october launch isnt that optimistic since if you studied the Thouroughbred core you would see that it was designed for the purpose of adding more cache.
Now everyone who disagrees with me. Get this through your thick heads. My point AXP+DDR vs. P4+RDRAM almost every benchmark comparing the two uses this combo. There are many DDR chipsets for the Athlon as well as many for the P4. We all know that there is 850 and 850E are the only two chipsets out right this very second for the p4. RDRAM offers a lot of memory bandwidth that the p4 can take advantage of unlike the Athlon can. DDR looks to be the offical standard for the next couple years and the overwelming majority of all PC's sold use DDR. (NOW here is the 2nd part of my argument) Someone who knows nothing about PC's wants to buy a brand new PC. He sees a PC with an Intel 845g and a 2.26ghz P4. A not uncommon setup for a OEM PC sold. Now it uses the Intel onboard graphics and DDR 266. He sees another PC an AthlonXP 2200+ with a Nvidia Nforce chipset DDR 266 which is not an uncommon setup. He doesnt know which to buy! Lets say they are equally priced as well if not the Athlon being cheaper. He looks online to see which computer is faster! He stumbles along on these hardware sites. HE sees the P4 crushing the AthlonXP in areas of memory bandwidth and games. He assumes the p4 is the best buy. Little does he know but his p4 doesnt use RDRAM which gives it the advantage in the various tests. Not only that but his Intel "Extreme Graphics" are eating his memory bandwidth as well just plain sucking. The benches he saw all used RDRAM 1066 which is not used in any p4 systems sold by major manufacturers. He should have picked the Nforce setup we all know that. This guy isn't a hardware entusiast he doesnt even know what a chipset is! <~~~~This guy is your average pc buyer and he makes up 97 percent of the market. Our small niche market doesnt really make a dent into te billions of dollars that the major PC companies make. We know what is good and he doesnt. HE sees his processor and another processor compared and thinks that is the only difference. HE doesnt know that your RAM or motherboard chipset can affect your processors performance.
p4 system performance right now is unequal with PC1066 and I for one would like to see it compared to the AthlonXP. Though I would also like to see what the overwhelming majority of PC buyers are getting their hands on. The value OEM pc for the most part sold is a much better deal with an Athlon setup. Thanks to the Nforce chipset and the KT333/266A. Why can't benches show p4+ddr, p4+PC1066, and AXP+ddr? If I said lets compare an Athlon with SDRAM vs a P4 with DDR most if not all people woluld say that was unfair. Well comparing an Athlon with DDR vs a P4 with PC1066 is not that much different.
AskHeart, Those XBit benchmarks are interesting.
In aceshardware, they also use PC2700 DDR CL2 with an i845g. In SpecViewPerf, the P4-2.53 handily beat the Athlon 2600+ on almost all fronts whereas it was the opposite in Xbit. Also, in every other review I've come across, there is a no competition from LightWave as the P4 being the champ.
I do not believe you are correct about Sandra FPU optimizations. First off, a Pentium III Tualatin at 512KB cache is a powerful performer. At 1.4Ghz, it can perform on par with a 1500+. Since Intel was touting its P4 platform, the Tualatin benchmarks are rare and far inbetween. The few that exist indicate that it does perform within 5% of an XP. Besides, if you look at Sandra closer, you'll see FPU and iSSE2. The latter being iSSE2 optimized.
My point was that in Sandra CPU bench, memory bandwidth has little to nothing to deal with the results that it generates, and thus real indication of performance based on the Sandra CPU bench is close to impossible to determine.
If consumers fail to educate themselves it is their fault. There is no information that is being hidden. The test setup apparatus are clearly marked. The hardware and software settings are clearly marked. Moreover, a P4 is designed with RDRam in mind that is synchronous to their FSB and an Athlon is designed with DDR SDRam that is also synchronous to their FSB. If you deviate from their design, I think you are more or less running it out of spec (DDR SDRam wasnt even intended for the p4 when the p4 was first introduced).
In aceshardware, they also use PC2700 DDR CL2 with an i845g. In SpecViewPerf, the P4-2.53 handily beat the Athlon 2600+ on almost all fronts whereas it was the opposite in Xbit. Also, in every other review I've come across, there is a no competition from LightWave as the P4 being the champ.
I do not believe you are correct about Sandra FPU optimizations. First off, a Pentium III Tualatin at 512KB cache is a powerful performer. At 1.4Ghz, it can perform on par with a 1500+. Since Intel was touting its P4 platform, the Tualatin benchmarks are rare and far inbetween. The few that exist indicate that it does perform within 5% of an XP. Besides, if you look at Sandra closer, you'll see FPU and iSSE2. The latter being iSSE2 optimized.
My point was that in Sandra CPU bench, memory bandwidth has little to nothing to deal with the results that it generates, and thus real indication of performance based on the Sandra CPU bench is close to impossible to determine.
If consumers fail to educate themselves it is their fault. There is no information that is being hidden. The test setup apparatus are clearly marked. The hardware and software settings are clearly marked. Moreover, a P4 is designed with RDRam in mind that is synchronous to their FSB and an Athlon is designed with DDR SDRam that is also synchronous to their FSB. If you deviate from their design, I think you are more or less running it out of spec (DDR SDRam wasnt even intended for the p4 when the p4 was first introduced).
Maj,
One question for you before I leave this thread. Would you object to reporting both RDRAM and DDR SDRAM scores for the P4 in comparisons with the XP?
Dexvx,
The reason the P4 tromps the XP in Lightware on sites like Toms Hardware is because they are using Intel recommended Lightwave tests. Lightwave is like Photoshop. It has lots of "filters". Some perform well on the P4 and some perform well on the P4. Guess which Intel recommends reviewers use... At Xbit labs they went beyond the Intel recommended settings and added their own tests on other parts of Lightwave. Thus the different scores. In the end these scores only matter if you personally use Lightwave in the manner it was tested. But it is a rather popular application among 3D artists.
One question for you before I leave this thread. Would you object to reporting both RDRAM and DDR SDRAM scores for the P4 in comparisons with the XP?
Dexvx,
The reason the P4 tromps the XP in Lightware on sites like Toms Hardware is because they are using Intel recommended Lightwave tests. Lightwave is like Photoshop. It has lots of "filters". Some perform well on the P4 and some perform well on the P4. Guess which Intel recommends reviewers use... At Xbit labs they went beyond the Intel recommended settings and added their own tests on other parts of Lightwave. Thus the different scores. In the end these scores only matter if you personally use Lightwave in the manner it was tested. But it is a rather popular application among 3D artists.
There is a simple way to proof my installment. Watch Sandra's whetstone x87 FPU scores for P4, now watch P4's SSE2 whetstone scores. P4's SSE2 and the x87 FPU share the same pipeline and Sandra's testing methodology is fairly equal to either x87 and SSE2 FPUs (no SSE2 operand vector-packing).
SSE2 scores higher than x87 because SSE2 enables a flat register file (no FPU stack),
The Tualatin is based off the P6 core, which doesn't suffer from the x87 FPU stack in the same manner Athlon and P4 FPU's do. The Tualatin is a strong core, but once the loading/storing of operands is optimized for the Athlon potent load/store unit, even the Tualatin is no match, when comparing the performance of the two per clock (potentially the Athlon's x87 FPU is +/- 70% stronger than Tualatin's x87 FPU per clock cycle).
Had Sandra specific optimizations for Athlon's or P4's x87 FPUs, they'd score higher, likely.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter