In light of THG's new a64 3400+ "review" (if you really think you can call it that) I've been disgusted by this site just once to much. I've decided to take the liberty of tearing apart this review and exposing what utter bullshit it is. Let's go page for page, shall we? For those of you looking for review, it is <A HREF="http://www.tomshardware.com/cpu/20040106/index.html" target="_new">here</A>. Let's show just how biased this "impartial" reviewer really is.
Page 1:
Page 2: Quite a worthless page here. We've got a pic of a 3000+ and a chopped up pic of a 3400+ stepping sticker. IMO this page was worthless and unneeded, but at least it isn't full of BS.
Page 3:
Page 4: Yay, they overclocked it, how exciting. Moving on...
Page 5: WTF is this bullshit!? The p4 setup gets ram with 2-2-2-5 timings, and nothing else does!? The <b>only</b> instance different ram should be used is the case of the FX-51, which requires registered ram. Otherwise the same ram should be used throught the entire test on ALL systems. C'mon, how biased can you get? I guess at the very least they were honest about the mem timings.
Page 6: Quake 3 is extremely Intel biased. The default DLL's don't even let AMD cpu's use SSE instructions. We're throwing this one out.
Page 7: SPECviewperf... wtf is this? I bet millions of people world-wide use this one on a daily basis. The first portion of the test shows the cpu's within a single frame per second of each other. Nothing to talk about, we're calling this one a draw. But the second part shows us a different story. The 3400+ is ahead of the 3.2c by more than 20fps. The third portion aslo shows us differences of less than a single frame per second. I find it rather fishy that the 3200+ is doing better than the 3400+ here. Maybe just a discrepancy, but it is a little strange. 3400+ - 1 3.2c - 0
Page 8: What the hell kind of benchmark has hardly more than a single fps difference between a 2500+ and a 3400+? But, I'm here to bitch about THG, not the useless benchmarks people use. 3400+ - 1 3.2c - 1
Page 9: Serious sam is more than 15 frames faster with the 3400+. A hands down victory. 3400+ - 2 3.2c - 1
Page 10: In ET the 3400+ has less than 2 fps difference with the 3.2c. I'm calling this one a draw.
Page 11: The 3400+ bests the 3.2c by nearly 1500 3dmarks. A clear victory here. 3400+ - 3 3.2c - 1
Page 12: About a 3 and a half fps difference here. Nothing to talk about IMO, we're calling it a draw.
Page 13: The 3400+ is well over 30fps faster, a clear victory. 3400+ - 4 3.2c - 1
Page 14: The 3400+ is nearly 15fps faster, score one for the 3400+. 3400+ - 5 3.2c - 1
Page 15: The difference here is a tenth of frame, I'm calling it a draw.
Page 16: Why in God's name would you use a gpu dependant benchmark to measure a cpu's performance!? The missing 100 points from the first half could easily be made up if THG had the brains or lack of bias to use the same ram on all the systems. I'm throwing the first half out. The second half is a little different. It's more cpu dependant (at least when using the same video card to compare) than the first half. The 3400+ has a decent lead, and one that would be larger if THG could've been fair and used the same ram. Another interesting thing to note here is that they didn't even use the latest version of 3dmark03. WTF is that!? I guess it doesn't really matter too much since nVidia's "optimizations" are present in both the 3400+ bench and the 3.2c bench. 3400+ - 6 3.2c - 1
Page 17: Can't say I've ever heard of this game, but I guess it's still there. The 3400+ has the lead by more than 15fps. 3400+ - 7 3.2c - 1
Page 18: The 3400+ is ahead by nearly 9fps here. Not a whole helluva lot, but I'm giving it to the 3400+.
3400+ - 8 3.2c - 1
Page 19: The results here are withing a single frame, I'm calling it a draw. An interesting thing to note here is the lack of default settings. This makes the results uncomparable to any default Aquamark3 benchmarks. A good reviewer should use default settings for a benchmark.
Page 20: I cannot tell a lie, Intel is the king of encoding. 3400+ - 8 3.2c - 2
Page 21: Intel is ahead here by roughly 2%. Not a clear victory IMO, but I'm giving it to them anyway.
3400+ - 8 3.2c - 3
Page 22: The 3.2c has a slight lead here as well, and again I'm giving them the win. 3400+ - 8 3.2c - 4
Page 23: Once again Intel shows its dominance in encoding.
3400+ - 8 3.2c - 5
Page 24: Again we see the 3.2c trouncing the 3400+ in encoding. 3400+ - 8 3.2c - 5
Page 25: Deja vuz. (sp?) 3400+ - 8 3.2c - 6
Page 26: Ya know, you really would think a single test would be enough to show that higher mhz = better encoding power. 3400+ - 8 3.2c - 7
Page 27: Another application that everyone uses in their everyday lives... Another for the 3.2c nonehteless
3400+ - 8 3.2c - 8
Page 28: Bapco is an intel biased POS benchmark and everyone knows it. I'm throwing this one out.
Page 29: Well, for the first time we see the 3.2c doing better than the 3400+ in a real life application that isn't encoding media. 3400+ - 8 3.2c - 9
Page 30: A minor lead for the 3400+. 3400+ - 9 3.2c - 9
Page 31: FFS, how many times do we have to show that the p4 is better for encoding? 3400+ - 9 - 3.2c - 10
Page 32: At last, another real life application. 2fps is nothing to talk about, IMO.
Page 33: The 3400+ absolutely trounces the 3.2c. 10 - 10
Page 34: PCMark is worthless and everyone knows it.
Page 35: How intelligent, benching the perfomance of memory on cpu's and not even using the same memory. Morons. I'm throwing this out.
Page 36:
Anyone that tells me THG isn't biased is beyond ignorant.
Page 1:
Umm... wtf is this [-peep-]!? The athlon 64 3200+ is <A HREF="http://www.newegg.com/app/ViewProductDesc.asp?description=19-116-164&catalog=343&depa=1" target="_new">$299 @ newegg</A>. The p4 3.2c on the other hand will set you back <A HREF="http://www.newegg.com/app/ViewProductDesc.asp?description=19-116-164&catalog=343&depa=1" target="_new">$397 @ newegg</A> I'll give the reviewer a little slack here, no doubt the prices have dropped on the 3200+ since the release of the 3400+. But even at $400, the a64 is still roughly the same price as Intel's competition. Don't tell me it costs too much, it's the same damn price as the competition! I do agree with the rest of the page, the A64 3000+ is an excellent chip for the money.Article Info
AMD Looks to Upstage Intel's P4 3.2 GHz Performance with Athlon64 3400+
--------------------------------------------------------------------------------
Created:
January 6, 2004
--------------------------------------------------------------------------------
By:
Patrick Schmid
Bert Töpelt
--------------------------------------------------------------------------------
Category:
Processors
--------------------------------------------------------------------------------
Summary:
The Athlon64 3400+ sure is fast, but can it beat the 3.2 GHz Pentium 4 performance-wise?
The Fours Have It: Athlon64 3400+ For 2004
The Athlon64 continues to gain in popularity since its September 2003 launch. One reason for this is that the processors are generally readily available; another, however, is that more and more manufacturers now make motherboards for the processor with Socket 754.
But that alone is not enough to justify the around $400 / €425 asking price for an Athlon64 3200+ - not exactly pocket change. AMD is aware of that fact and responded a few weeks ago with the Athlon64 3000+. It's a 3200+ with a pared down 512 KB L2 cache and the same 2.0 GHz clock speed. At $225 / €250, the price is the only thing they skimped on.
Page 2: Quite a worthless page here. We've got a pic of a 3000+ and a chopped up pic of a 3400+ stepping sticker. IMO this page was worthless and unneeded, but at least it isn't full of BS.
Page 3:
Right... 89 watts is the <b>peak output</b> of all A64's up to the 3700+. 95 watts is a made-up number THG pulled out of their asses. The rest of the article just seems to babble on about technology that IMO is utterly worthless for a desktop cpu. I can see where it would relevant for a laptop cpu... but this is just stupid IMO. As far as I can tell though, the rest of the article is factual.How to keep the processor cool without ratcheting down performance remains an issue. With the Athlon64 3200+ you're talking about up to 89 watts being converted into thermal energy; the 3400+ will likely put out around 95 watts.
Page 4: Yay, they overclocked it, how exciting. Moving on...
Page 5: WTF is this bullshit!? The p4 setup gets ram with 2-2-2-5 timings, and nothing else does!? The <b>only</b> instance different ram should be used is the case of the FX-51, which requires registered ram. Otherwise the same ram should be used throught the entire test on ALL systems. C'mon, how biased can you get? I guess at the very least they were honest about the mem timings.
Page 6: Quake 3 is extremely Intel biased. The default DLL's don't even let AMD cpu's use SSE instructions. We're throwing this one out.
Page 7: SPECviewperf... wtf is this? I bet millions of people world-wide use this one on a daily basis. The first portion of the test shows the cpu's within a single frame per second of each other. Nothing to talk about, we're calling this one a draw. But the second part shows us a different story. The 3400+ is ahead of the 3.2c by more than 20fps. The third portion aslo shows us differences of less than a single frame per second. I find it rather fishy that the 3200+ is doing better than the 3400+ here. Maybe just a discrepancy, but it is a little strange. 3400+ - 1 3.2c - 0
Page 8: What the hell kind of benchmark has hardly more than a single fps difference between a 2500+ and a 3400+? But, I'm here to bitch about THG, not the useless benchmarks people use. 3400+ - 1 3.2c - 1
Page 9: Serious sam is more than 15 frames faster with the 3400+. A hands down victory. 3400+ - 2 3.2c - 1
Page 10: In ET the 3400+ has less than 2 fps difference with the 3.2c. I'm calling this one a draw.
Page 11: The 3400+ bests the 3.2c by nearly 1500 3dmarks. A clear victory here. 3400+ - 3 3.2c - 1
Page 12: About a 3 and a half fps difference here. Nothing to talk about IMO, we're calling it a draw.
Page 13: The 3400+ is well over 30fps faster, a clear victory. 3400+ - 4 3.2c - 1
Page 14: The 3400+ is nearly 15fps faster, score one for the 3400+. 3400+ - 5 3.2c - 1
Page 15: The difference here is a tenth of frame, I'm calling it a draw.
Page 16: Why in God's name would you use a gpu dependant benchmark to measure a cpu's performance!? The missing 100 points from the first half could easily be made up if THG had the brains or lack of bias to use the same ram on all the systems. I'm throwing the first half out. The second half is a little different. It's more cpu dependant (at least when using the same video card to compare) than the first half. The 3400+ has a decent lead, and one that would be larger if THG could've been fair and used the same ram. Another interesting thing to note here is that they didn't even use the latest version of 3dmark03. WTF is that!? I guess it doesn't really matter too much since nVidia's "optimizations" are present in both the 3400+ bench and the 3.2c bench. 3400+ - 6 3.2c - 1
Page 17: Can't say I've ever heard of this game, but I guess it's still there. The 3400+ has the lead by more than 15fps. 3400+ - 7 3.2c - 1
Page 18: The 3400+ is ahead by nearly 9fps here. Not a whole helluva lot, but I'm giving it to the 3400+.
3400+ - 8 3.2c - 1
Page 19: The results here are withing a single frame, I'm calling it a draw. An interesting thing to note here is the lack of default settings. This makes the results uncomparable to any default Aquamark3 benchmarks. A good reviewer should use default settings for a benchmark.
Page 20: I cannot tell a lie, Intel is the king of encoding. 3400+ - 8 3.2c - 2
Page 21: Intel is ahead here by roughly 2%. Not a clear victory IMO, but I'm giving it to them anyway.
3400+ - 8 3.2c - 3
Page 22: The 3.2c has a slight lead here as well, and again I'm giving them the win. 3400+ - 8 3.2c - 4
Page 23: Once again Intel shows its dominance in encoding.
3400+ - 8 3.2c - 5
Page 24: Again we see the 3.2c trouncing the 3400+ in encoding. 3400+ - 8 3.2c - 5
Page 25: Deja vuz. (sp?) 3400+ - 8 3.2c - 6
Page 26: Ya know, you really would think a single test would be enough to show that higher mhz = better encoding power. 3400+ - 8 3.2c - 7
Page 27: Another application that everyone uses in their everyday lives... Another for the 3.2c nonehteless
3400+ - 8 3.2c - 8
Page 28: Bapco is an intel biased POS benchmark and everyone knows it. I'm throwing this one out.
Page 29: Well, for the first time we see the 3.2c doing better than the 3400+ in a real life application that isn't encoding media. 3400+ - 8 3.2c - 9
Page 30: A minor lead for the 3400+. 3400+ - 9 3.2c - 9
Page 31: FFS, how many times do we have to show that the p4 is better for encoding? 3400+ - 9 - 3.2c - 10
Page 32: At last, another real life application. 2fps is nothing to talk about, IMO.
Page 33: The 3400+ absolutely trounces the 3.2c. 10 - 10
Page 34: PCMark is worthless and everyone knows it.
Page 35: How intelligent, benching the perfomance of memory on cpu's and not even using the same memory. Morons. I'm throwing this out.
Page 36:
Fucktards. Lets have a look now. ^^^ It seems Intel won 10/20 benchmarks with any value whatsoever. And with only 1 exception, these were all media encoding.A close look at all the benchmark results reveals that the new Athlon64 just barely earns the performance rating 3400+. Out of 32 benchmarks, only 13 were decided clearly in favor of AMD's new contender. If you were to evaluate each of the 41 individual disciplines, the result would be even poorer.
Lags behing the p4? Whatever. Have a look. The ONLY thing the p4 does better is encoding.At the end of the day, it still lags slightly behind the Pentium 4, a deficit that the 64-bit architecture could compensate for in the medium term, however.
Anyone that tells me THG isn't biased is beyond ignorant.