Alright, I wrote in one other time, and I believed that a certain review that I read was ridiculous. I realize that, while it was ridiculous, it was mostly a subjective feeling, rather than a fact.
However, I read a review today that I believe to not only be horribly incompetent, but flat out misleading. The review is from MBReview.com, and the specific review I'm speaking about is this one.
Now, this review is truly incredible. First is some general information about DDR and how it transmits data on the top and bottom of each clock cycle, which is why a stick running at say 200MHz is running at and effective rate of 400MHz DDR. We'll refer back to "true" MHz versus "effective" MHz in a minute.
The next thing he mentions is that he couldn't get any stick to above 410 MHz DDR. This will also be important in a minute.
Now we have some Sandra benchmarks, and obviously enough, all the sticks perform exactly the same within a negligible statistical variance. He states that, "at the 166MHz test speed, the XtremeDDR PC-3500+ comes out the winner for integer performance, while the Corsair RAM takes the floating-point performance." What in the hell is he smoking? The scores for the two tests and the two sticks are 2040 vs. 2037 and 1944 vs. 1950, respectively. How in the hell can he claim a winner in that battle?
The next two pages worth of tests are far better (it's so hard to type sarcasm). If you don't see this for yourself, you won't believe it. The first page is a couple of Quake 3 benchmarks that have frames per second out to the tenth of a frame! Oh, but that's not all. Now we go to the next page with UT2003 benches. These are taken out to the thousandth of a frame! The largest variation of any one stick to another on this page is half of a frame, and most don't vary by much more than 2/10 of a frame! Seriously, I had to make sure that this wasn't a joke more than once. Just read what he says based on less than half a frame's difference:
Just in case you thought the madness stopped there, you're wrong. The next page is full of results from overclocking. First, he says that Corsair XMS gets to "204MHz DDR... @ 2.5CL." Then he says that he got the Samsung to "203MHz DDR stably," but does not list the CL rating. I would assume that if it was true Samsung, it usually O/C's the same or better than respective Corsair XMS memory. So since the Samsung was PC2700 and the Corsair was PC3000, I would say that the Samsung was probably at CL2.5 as well, but maybe even CL2. Now, at this point, he has labeled both sticks at their "true" MHz rating, rather than their DDR rating. He's already says that he knows the difference, so this cannot simply be ignorance.
Now, we get to the RAM being reviewed: XtremeDDR PC-3500+. This "show-stopper of the night" as he calls it gets to "a cool 410MHz DDR @ CL3." Woah... there are so many problems with that statement, I have to take a second to get all of my complaints in order. First off, why in the hell did he switch to using the DDR rating of the RAM? If my mom read this (and she knows a little about how computers work), she would think that this RAM was able to run over twice as fast as the other two sticks. However, unless they were able to reach a SCREAMING 820MHz DDR!!!!, then this is just misleading. In fact, we all know that that's just not possible. Next, look at the CL rating. 3? I'll tell you this: if you drop the CL rating of a stick of RAM, you had better expect more than 1 MHz increase in clock speed. Then he goes on to say that this Editor's-Choice-worthy stick can't even get to it's rated speed! But hey, this is the board's fault. Remember, he states here the same thing that he said before: the board won't go faster than 205MHz. (In my head, I'm yelling this part) Well if the board is limiting your review, then how can you give an overclocking award to any of these sticks of RAM?!?!
So here are my gripes laid out:
- He talks like he knows something about RAM, so he can't claim pure ignorance
- He uses ridiculous benchmarks that show microscopic differences in scores and calls one stick the winner over the other.
- He incorrectly labels the overclock speeds of the first two sticks of RAM (should be just MHz, not MHz DDR)
- He compares the overclock of one stick of RAM to another at different CL ratings
- He says it's OK to not hit the speed the RAM is rated for, blaming it on the board (another gripe)
Please, all AT readers, only go to this site long enough to laugh your arse off at this ridiculous review, and then never return. If we can decrease the number of hits that this site gets, hopefully it will go the way of the Dodo. While this site definitely pisses me off, it also makes me bust up laughing. Enjoy... only once.
However, I read a review today that I believe to not only be horribly incompetent, but flat out misleading. The review is from MBReview.com, and the specific review I'm speaking about is this one.
Now, this review is truly incredible. First is some general information about DDR and how it transmits data on the top and bottom of each clock cycle, which is why a stick running at say 200MHz is running at and effective rate of 400MHz DDR. We'll refer back to "true" MHz versus "effective" MHz in a minute.
The next thing he mentions is that he couldn't get any stick to above 410 MHz DDR. This will also be important in a minute.
Now we have some Sandra benchmarks, and obviously enough, all the sticks perform exactly the same within a negligible statistical variance. He states that, "at the 166MHz test speed, the XtremeDDR PC-3500+ comes out the winner for integer performance, while the Corsair RAM takes the floating-point performance." What in the hell is he smoking? The scores for the two tests and the two sticks are 2040 vs. 2037 and 1944 vs. 1950, respectively. How in the hell can he claim a winner in that battle?
The next two pages worth of tests are far better (it's so hard to type sarcasm). If you don't see this for yourself, you won't believe it. The first page is a couple of Quake 3 benchmarks that have frames per second out to the tenth of a frame! Oh, but that's not all. Now we go to the next page with UT2003 benches. These are taken out to the thousandth of a frame! The largest variation of any one stick to another on this page is half of a frame, and most don't vary by much more than 2/10 of a frame! Seriously, I had to make sure that this wasn't a joke more than once. Just read what he says based on less than half a frame's difference:
Based on half a frame!!!!!!
Just in case you thought the madness stopped there, you're wrong. The next page is full of results from overclocking. First, he says that Corsair XMS gets to "204MHz DDR... @ 2.5CL." Then he says that he got the Samsung to "203MHz DDR stably," but does not list the CL rating. I would assume that if it was true Samsung, it usually O/C's the same or better than respective Corsair XMS memory. So since the Samsung was PC2700 and the Corsair was PC3000, I would say that the Samsung was probably at CL2.5 as well, but maybe even CL2. Now, at this point, he has labeled both sticks at their "true" MHz rating, rather than their DDR rating. He's already says that he knows the difference, so this cannot simply be ignorance.
Now, we get to the RAM being reviewed: XtremeDDR PC-3500+. This "show-stopper of the night" as he calls it gets to "a cool 410MHz DDR @ CL3." Woah... there are so many problems with that statement, I have to take a second to get all of my complaints in order. First off, why in the hell did he switch to using the DDR rating of the RAM? If my mom read this (and she knows a little about how computers work), she would think that this RAM was able to run over twice as fast as the other two sticks. However, unless they were able to reach a SCREAMING 820MHz DDR!!!!, then this is just misleading. In fact, we all know that that's just not possible. Next, look at the CL rating. 3? I'll tell you this: if you drop the CL rating of a stick of RAM, you had better expect more than 1 MHz increase in clock speed. Then he goes on to say that this Editor's-Choice-worthy stick can't even get to it's rated speed! But hey, this is the board's fault. Remember, he states here the same thing that he said before: the board won't go faster than 205MHz. (In my head, I'm yelling this part) Well if the board is limiting your review, then how can you give an overclocking award to any of these sticks of RAM?!?!
So here are my gripes laid out:
- He talks like he knows something about RAM, so he can't claim pure ignorance
- He uses ridiculous benchmarks that show microscopic differences in scores and calls one stick the winner over the other.
- He incorrectly labels the overclock speeds of the first two sticks of RAM (should be just MHz, not MHz DDR)
- He compares the overclock of one stick of RAM to another at different CL ratings
- He says it's OK to not hit the speed the RAM is rated for, blaming it on the board (another gripe)
Please, all AT readers, only go to this site long enough to laugh your arse off at this ridiculous review, and then never return. If we can decrease the number of hits that this site gets, hopefully it will go the way of the Dodo. While this site definitely pisses me off, it also makes me bust up laughing. Enjoy... only once.
