The benchmarks show nothing if you don't understand what it is they are testing. SR's benchmarks only test for peak peformance under ideal conditions. Their results won't show the effect of real world performance issues like file system fragmentation and using the slower parts of the drive that anyone who actually uses a hard drive would experience. They use IOmeter to test server performance. If you know how IOmeter tests drive performance you know why platter count is completely irrelevant to it, and why SR has stopped using it for workstation (single user) benchmarking. Their workstation benchmarks again are testing ideal scenario performance. How much space do you think their OS installation and benchmarks take up? 10GB tops? How does using that much space on a 400GB vs 250GB drive where more than 95% of each drive is empty and all data is perfectly fragmented going to mimic real world usage that would show the affects of adding platters? If they could partition the drive and test each section of the drive, I guarantee you you would see noticeable differences in performance. This still wouldn't take into account fragmentation and having to seek over different portions of the platter which is practically impossible to replicate. "Real world" hard drive benchmarks that are easily repeatable are practically impossible to derive which is why all we have are mostly lowlevel benchmarks which can't tell the whole story.
The benchmarks show nothing if you don't understand what it is they are testing. SR's benchmarks only test for peak peformance under ideal conditions. Their results won't show the effect of real world performance issues like file system fragmentation and using the slower parts of the drive that anyone who actually uses a hard drive would experience. They use IOmeter to test server performance. If you know how IOmeter tests drive performance you know why platter count is completely irrelevant to it, and why SR has stopped using it for workstation (single user) benchmarking. Their workstation benchmarks again are testing ideal scenario performance. How much space do you think their OS installation and benchmarks take up? 10GB tops? How does using that much space on a 400GB vs 250GB drive where more than 95% of each drive is empty and all data is perfectly fragmented going to mimic real world usage that would show the affects of adding platters? If they could partition the drive and test each section of the drive, I guarantee you you would see noticeable differences in performance. This still wouldn't take into account fragmentation and having to seek over different portions of the platter which is practically impossible to replicate. "Real world" hard drive benchmarks that are easily repeatable are practically impossible to derive which is why all we have are mostly lowlevel benchmarks which can't tell the whole story.
I'm gonna try one more time 
I'm agreeing with you, BUT
1 The OP I assumed was keeping consistent in his data quantity, 2 The article you pulled off storage review was from 2001, my information 2004. You article outright says in its conclusions it doesn't have the wherewithall to test real world data volumes back then. In the situation of equal amounts of data his bottome line is this:
"The higher capacity designs provide better transfer rates as a function of GBs-used-on-the-drive (but not % of capacity) because of their higher data density"
So when he asks is bigger faster? if its in the same drive family and if the quantity of data is the same then yes it is.
By roughly the percentages I gave, If its in % of drive space then the smaller drive wins !! and faster by as much as 10%
"However, you can usually count on the lower-capacity designs to perform better in workloads that involve predominantly random seeks because they usually have faster move times."
So the OP just has to ask what kind of computing is he doing? if he is in a two drive system already I wouldn't bother partitioning. Personally I prefer to get the fastest drive technology as I said, slowest computer component rather than worry bout these scenarios
I'm agreeing with you, BUT
1 The OP I assumed was keeping consistent in his data quantity, 2 The article you pulled off storage review was from 2001, my information 2004. You article outright says in its conclusions it doesn't have the wherewithall to test real world data volumes back then. In the situation of equal amounts of data his bottome line is this:
"The higher capacity designs provide better transfer rates as a function of GBs-used-on-the-drive (but not % of capacity) because of their higher data density"
So when he asks is bigger faster? if its in the same drive family and if the quantity of data is the same then yes it is.
By roughly the percentages I gave, If its in % of drive space then the smaller drive wins !! and faster by as much as 10%
"However, you can usually count on the lower-capacity designs to perform better in workloads that involve predominantly random seeks because they usually have faster move times."
So the OP just has to ask what kind of computing is he doing? if he is in a two drive system already I wouldn't bother partitioning. Personally I prefer to get the fastest drive technology as I said, slowest computer component rather than worry bout these scenarios
No, SR's test suite doesn't only use 10GB. Besides testing to the end of the drive, it requires over 30GB of space, which is why they weren't able to go back and benchmark a lot of older drives.
Anyway, you're saying that StorageReview's benchmarks are irrelevant, yet you have nothing but your own conjecture to refute them. Do you have any idea how ridiculous that sounds? Maybe SR's testbed doesn't tell the whole story, but do you have anything better to offer besides random speculation?
It's not even good speculation. You're talking about things which have been known about for years, and those in the industry agree that those factors have minimal impact.
Why don't you just go email Eugene and explain what a genius you are? Tell him about your uncanny ability to understand things which you haven't even tested. He'll probably offer you a position on SR. Hell, he might even step down and hand it all over to you since you've done such a bang-up job of disproving him.
Feel free to check out some benchmarks:
Raptor 36 vs. modern 7200RPM drives
The 7200RPM drives are just as fast as the Raptor 36. They're actually faster in everything except the server performance benchmarks.
Now look at the price:
Raptor 36 = $110
A Raptor 36 is a complete waste of money because for that price you can get a 200GB drive that is just as fast.
The test image takes up 30GB exactly, which is more than I thought, but doesn't actually change anything, it's still too small a size to illustrate the issue at hand with today's capacity monsters. Look at the STR graph for the 7K250. It doesn't start to really decay until after the 60GB point. By increasing the platter count, not much is going to change since the drives are both running in the same zone. Going from 250GB to 400GB is only a 40% increase is platter surface while going from 80GB to 250 is a 200% increase. Which comparison do you think is going to show the bigger difference?
I didn't say SR's benchmarks are irrelevant. I said if you don't understand what they are testing and how those methods apply to real world scenarios they mean nothing. That goes for any benchmark, not just SR's.
I've had many conversations with Eugene, and for the record, he is far more pleasant to converse with than you.
I didn't say SR's benchmarks are irrelevant. I said if you don't understand what they are testing and how those methods apply to real world scenarios they mean nothing. That goes for any benchmark, not just SR's.
I've had many conversations with Eugene, and for the record, he is far more pleasant to converse with than you.
*crying* I'm still indenial. I still think their fast. But looks like benchmarks are benchmarks. Thanks for the link!
If that's true, why don't you have another conversation with him? Ask him if you should be influencing someone's buying decision with speculation about transfer rate discrepancies between 80GB and 200GB models of the same drive. I'll guarantee he'll tell you the same thing I am -- that the drives will perform similarly, and even if the 200GB model did have a slight edge, it would not be big enough to be a factor in deciding which hard drive this guy should buy.
Hey, if you have a Raptor you'd like to get rid of....
I'm just saying if you're going out and buying a new drive, a Raptor 36 really isn't cost effective anymore. Obviously it's still a great drive.
No, i'm keeping mine. I have 2 in RAID 0 right now. and they speed things up a bit. I have been considering a new HD and was gonna go with 74gb raptor, but it's looking like another larger drive would be best bang for buck.
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 + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter