Okay, I'm sick of a certain few people who think it's the end of the world that A64 didn't win a fight it should have.
So I'm going to try to put an end to all of this complaining...
1) Yes, THG did not use fair timings for the A64...get over it, they may have had a bad board or RAM for the A64--it's not like they pay for it.
2) Yes, THG was wrong about the pricing of A64. Turns out the article was written well in advance of the price change, how would they have known the 3200+'s price would plummet? They can't...get over it.
3) Yes, the 3400+ was deemed "barely a 3400+." Well, they are sort of right, mathematically speaking at least...more on this later.
4) Yes, they ran seemingly pointless synthetic benchmarks...but so does everyone else. Here's why: if a program is not benchmarked, the closest way to approximate it's performance is through a synthetic benchmark. For better or worse, one of these benchmarks was Sandra...which is very clockspeed and bandwidth dependent, giving a huge advantage to the P4C. They have been ignored in the statistical analysis of this post.
5) There are more factors to whether or not a CPU is a wise investment other than benchmarks. One feature that is will really start to become a prominent and important (IMO, at least--reasons following) is Hyper Threading. First of all, it helps daily usage, helps multitasking tremendously, and more programs are starting to be coded for SMP, meaning they will go faster than without it.
Now, aren't you all wondering about the 3400+ business? I know you are. Okay, I made a spreadsheet of all benchmark scores of the 3400+ and the 3.2. From there I found the percentage difference, if the percentage were positive, AMD had that much advantage, if the percentage were negative, AMD had that much of a disadvantage (different than Intel having X% of an advantage). For each category (1. Gaming, 2. Application/synthetic, and 3. Media/rendering/compression) I totaled the percentages and averaged (yes, that is mathematically legal). I then totaled all 41 percentages and averaged, giving me four final percentages. One was an overall score (each benchmark was weighted equally), one was a media/rendering/compression score, one was a gaming score, and the last was an application/synthetic score.
According to the equation: 3400/3200, the 3400+ should be 6.25% faster than the 3.2C. If only this were the case...
RESULTS:
GAMING: 4.444% Advantage to AMD
MEDIA/RENDERING/COMPRESSION: 8.567% Disadvantage to AMD
<b>OVERALL: .682% DISADVANTAGE to AMD</b>
I am willing to email ANYBODY the .xls file where all this information is contained. Please note that ALL benchmarks were weighted equally.
According to my findings, the 3400+ should be named: 3179+
HOWEVER!!! THG's tests were not fairly run (mainly timings), so take this with a grain of salt...this adjustment will put the 3400+ on top (another grain of salt can be taken for me weighting all the benchmarks equally, but it's not the 6.25% it needs to fit it's name quota [granted that AMD fit the PR rating around Intel's MHz]).
Needless to say, THG does need to clean up its procedure for testing. As for conclusions, maybe they should be fairer to AMD (i.e, less biased), but if they do anything at all similar to me (i.e., a mathematical determination of the winner), they are not incorrect in saying the 3400+ is not a 3400+ (taking for granted the PR rating is meant to match the C's MHz). As for the winner in terms of current 32-bit performance, there is some discrepancy here, mainly depending on weighting of benchmarks and features.
Please note that in a few months when 64-bit finally becomes mainstream (i.e., XP 64-bit), AMD's advantages will most likely be unprecedented.
Also, since the majority of readers of THG (I presume) are gamers, THG is even morely (is that even a word?) wrong about putting down the 3400+ as the weaker chip, since it undeniably beats the 3.2C in gaming (but still not to it's 6.25% quota).
Oh yeah, I'm not just some Intelliot posting this to shut up everyone who is complaining, I fully respect and appreciate both company's contribution to each other and the market in general. In fact, if I had to build a system for myself today, it would be a dual Opteron system (yes, dual Xeon would be faster at the majority of what I do, but Opterons have more of a future for upgrade) or a highly OCed FX (if the dual Opterons can't be OCed).
Additionally, AMD has roadmapped a socket 754 3700+, meaning that there is room for upgrade further down the road. The likelyhood of a considerable (this sentence is contingent on the unknown performance of Scotty) upgrade for socket 478 is unlikely.
I hope that settles things.
P.S., can someone tell me the point of complaining about THG saying the 3400+ isn't up-to-snuff? It's not like the outcome hurts you in anyway or that they will change their conclusion. All it does is waste all of our time (especially for those who need help with a problem), nobody is advantaged by your complaining. I hope this post finally puts an end to all of this.
Maxtor disgraces the six letters that make Matrox.<P ID="edit"><FONT SIZE=-1><EM>Edited by Vapor on 01/07/04 01:52 PM.</EM></FONT></P>
So I'm going to try to put an end to all of this complaining...
1) Yes, THG did not use fair timings for the A64...get over it, they may have had a bad board or RAM for the A64--it's not like they pay for it.
2) Yes, THG was wrong about the pricing of A64. Turns out the article was written well in advance of the price change, how would they have known the 3200+'s price would plummet? They can't...get over it.
3) Yes, the 3400+ was deemed "barely a 3400+." Well, they are sort of right, mathematically speaking at least...more on this later.
4) Yes, they ran seemingly pointless synthetic benchmarks...but so does everyone else. Here's why: if a program is not benchmarked, the closest way to approximate it's performance is through a synthetic benchmark. For better or worse, one of these benchmarks was Sandra...which is very clockspeed and bandwidth dependent, giving a huge advantage to the P4C. They have been ignored in the statistical analysis of this post.
5) There are more factors to whether or not a CPU is a wise investment other than benchmarks. One feature that is will really start to become a prominent and important (IMO, at least--reasons following) is Hyper Threading. First of all, it helps daily usage, helps multitasking tremendously, and more programs are starting to be coded for SMP, meaning they will go faster than without it.
Now, aren't you all wondering about the 3400+ business? I know you are. Okay, I made a spreadsheet of all benchmark scores of the 3400+ and the 3.2. From there I found the percentage difference, if the percentage were positive, AMD had that much advantage, if the percentage were negative, AMD had that much of a disadvantage (different than Intel having X% of an advantage). For each category (1. Gaming, 2. Application/synthetic, and 3. Media/rendering/compression) I totaled the percentages and averaged (yes, that is mathematically legal). I then totaled all 41 percentages and averaged, giving me four final percentages. One was an overall score (each benchmark was weighted equally), one was a media/rendering/compression score, one was a gaming score, and the last was an application/synthetic score.
According to the equation: 3400/3200, the 3400+ should be 6.25% faster than the 3.2C. If only this were the case...
RESULTS:
GAMING: 4.444% Advantage to AMD
MEDIA/RENDERING/COMPRESSION: 8.567% Disadvantage to AMD
<b>OVERALL: .682% DISADVANTAGE to AMD</b>
I am willing to email ANYBODY the .xls file where all this information is contained. Please note that ALL benchmarks were weighted equally.
According to my findings, the 3400+ should be named: 3179+
HOWEVER!!! THG's tests were not fairly run (mainly timings), so take this with a grain of salt...this adjustment will put the 3400+ on top (another grain of salt can be taken for me weighting all the benchmarks equally, but it's not the 6.25% it needs to fit it's name quota [granted that AMD fit the PR rating around Intel's MHz]).
Needless to say, THG does need to clean up its procedure for testing. As for conclusions, maybe they should be fairer to AMD (i.e, less biased), but if they do anything at all similar to me (i.e., a mathematical determination of the winner), they are not incorrect in saying the 3400+ is not a 3400+ (taking for granted the PR rating is meant to match the C's MHz). As for the winner in terms of current 32-bit performance, there is some discrepancy here, mainly depending on weighting of benchmarks and features.
Please note that in a few months when 64-bit finally becomes mainstream (i.e., XP 64-bit), AMD's advantages will most likely be unprecedented.
Also, since the majority of readers of THG (I presume) are gamers, THG is even morely (is that even a word?) wrong about putting down the 3400+ as the weaker chip, since it undeniably beats the 3.2C in gaming (but still not to it's 6.25% quota).
Oh yeah, I'm not just some Intelliot posting this to shut up everyone who is complaining, I fully respect and appreciate both company's contribution to each other and the market in general. In fact, if I had to build a system for myself today, it would be a dual Opteron system (yes, dual Xeon would be faster at the majority of what I do, but Opterons have more of a future for upgrade) or a highly OCed FX (if the dual Opterons can't be OCed).
Additionally, AMD has roadmapped a socket 754 3700+, meaning that there is room for upgrade further down the road. The likelyhood of a considerable (this sentence is contingent on the unknown performance of Scotty) upgrade for socket 478 is unlikely.
I hope that settles things.
P.S., can someone tell me the point of complaining about THG saying the 3400+ isn't up-to-snuff? It's not like the outcome hurts you in anyway or that they will change their conclusion. All it does is waste all of our time (especially for those who need help with a problem), nobody is advantaged by your complaining. I hope this post finally puts an end to all of this.
Maxtor disgraces the six letters that make Matrox.<P ID="edit"><FONT SIZE=-1><EM>Edited by Vapor on 01/07/04 01:52 PM.</EM></FONT></P>