Archived

This topic is now archived and is closed to further replies.

Noel

Opinions/facts on how HDD performance affects FSX performance

Recommended Posts

My 36gb 15K rpm SCSI that FSX alone sits in is now almost full. I have a decision to make: clone this drive to another 73GB 15K SCSI I have, or dump it on a 250 gb SATA II 7200 rpm drive. It would appeare access time is about the only advantage with the 15K SCSI versus the 7200 drive has. Buffered read rate is actually much better with SATA II. So, I wonder, has anyone verified what benefits one can achieve with a better performing HDD, but more importantly, what HDD perf parameter is likely to be meaningful, if at all? I did clone my existing Vista 64 FSX installation to the 250gb SATA II, and don't think I'm noticing a difference. It's still quite fluid and I don't see to be getting lags at all. Here is an Everest comparison of the two drives:I wouldn't expect anything in terms of frame rate impact, but perhaps texture loading, etc. Again, not sure what matters more here: buffered read rate, random, or access time? I wouldn't mind losing the SCSI controller etc as it slows boot time. I purchased these 15K rpm scsi's to optimize GigaStudio's realtime disk streaming technology. It turns out the SATA II works fine in that environment too.Thanks for your expertise . . .Noel

Share this post


Link to post
Share on other sites
Help AVSIM continue to serve you!
Please donate today!

My 36gb 15K rpm SCSI that FSX alone sits in is now almost full. I have a decision to make: clone this drive to another 73GB 15K SCSI I have, or dump it on a 250 gb SATA II 7200 rpm drive. It would appeare access time is about the only advantage with the 15K SCSI versus the 7200 drive has. Buffered read rate is actually much better with SATA II. So, I wonder, has anyone verified what benefits one can achieve with a better performing HDD, but more importantly, what HDD perf parameter is likely to be meaningful, if at all? Noel
Just to complicate things... another factor is how far you allow a drive to fill up. More than 50% full has a negative performance impact,and I can imagine that a lower utilization may well have a positive impact since the disk arm has less distance to travel.

Share this post


Link to post
Share on other sites

Noel,Spend the money and get yourself a 300GB Western Digital Velociraptor for your FSX drive you will not be disappointed (you will have plenty of room for addons). You are keeping your FSX OS and FSX install on seperate drives are you not?

Share this post


Link to post
Share on other sites
Noel,Spend the money and get yourself a 300GB Western Digital Velociraptor for your FSX drive you will not be disappointed (you will have plenty of room for addons). You are keeping your FSX OS and FSX install on seperate drives are you not?
Yes I have always kept OS and FSX on separate drives. Sarg do you know what if anything matters in FSX with re to HD performance characteristics? I don't think I want to buy another drive right now because I have the two drives here I can play with. If one fails, then yes for sure I will pick up the racey drive you mentioned as I know it's always the top pick. What I was hoping to find out here was why it matters. You can see burst read is really ok with the lowly SATA II I have. Do you know what parameters affect anything meaningful with re to FSX? Access time seems kinda iffy really: 9ms faster access with my 15K SCSI, but I am thinking FSX request scenery loading in anticipation of where the sim is going, and if it's .9ms later, will this translate to anything meaningful (ie, percievable)? I would think CPU utilization could matter some. You can see, there is low CPU use with both drives. I appreciate the recommendation, but I still haven't heard from anyone with regard to what aspects of FSX perf are affected by HD perf characteristics.

Share this post


Link to post
Share on other sites
I appreciate the recommendation, but I still haven't heard from anyone with regard to what aspects of FSX perf are affected by HD perf characteristics.
Noel,And I expect you are not really going to hear from anyone who has hard data on the subject. Conventional wisdom would suggest that the drive with the fastest access and read times is going to provided better performance for FSX provided that the system to which the drive is connected can receive the data optimally. Synthetic benchmarks are not going to shed much light on real world or app specific performance. If your current 250GB SATA gives you satisfactory performance in FSX then by all means stick with it. However, it seems to me that you enjoy getting every little bit that you can out of FSX. :(

Share this post


Link to post
Share on other sites

From a more common sense point of view, one might consider that a hard drive's job is to transfer data to (and from) the system. That's the long of it . . . and the short of it. The rate at which the HD can transfer data to and from the system (ie data rate) will be the OnLy measurement parameter that will matter to a final performance result. For instance, the data must be accesses - before - it can be transferred to the system. Therefore, decreased access time (most certainly) has the potential to Increase data transfer rates, but as a stand-alone metric, it is meaningless. This is very similar to the ram speed discussion. Internal ram latency decreases only matter if they assist by Increasing the amount of data that can be transferred from/to the system. In that case, the controlling metric was memory buss <> system buss latency through the memory controller. Ram module internal timings (5-5-5-12/etc) and speeds (DDR12-50gazillion/etc) were meaningless unless they decreased memory buss <> system buss latency. As we remember, huge decreases in internal module latencies and equally huge increases in ram "speed" were required to produce decreases in this latency metric . . . and it took equally huge decreases in this latency metric (2X) to produce only arguable increases in FS performance. Great fun for the computer hobbiest, but a high risk/limited reward proposition for the FS-er. I ran some FS tests with HD transfer rates of ~ 250MB/s then with at 100MB/s. There was no perceivable ingame result and load times were only marginally affected. Overall, my system "felt" more responsive with the 250MB/s transfer configuration. but this was sensed only because one tends to develop a very tactile sense about their own system. Again, the FS change was indecipherable. These test indicate that HD transfer rates at ~ 100MB/s are more than FS needs for any aspect ingame play. Faster transfer rates might help game/flight load times, but by only a few seconds, at best.

Share this post


Link to post
Share on other sites
From a more common sense point of view, one might consider that a hard drive's job is to transfer data to (and from) the system. That's the long of it . . . and the short of it. The rate at which the HD can transfer data to and from the system (ie data rate) will be the OnLy measurement parameter that will matter to a final performance result. For instance, the data must be accesses - before - it can be transferred to the system. Therefore, decreased access time (most certainly) has the potential to Increase data transfer rates, but as a stand-alone metric, it is meaningless. This is very similar to the ram speed discussion. Internal ram latency decreases only matter if they assist by Increasing the amount of data that can be transferred from/to the system. In that case, the controlling metric was memory buss <> system buss latency through the memory controller. Ram module internal timings (5-5-5-12/etc) and speeds (DDR12-50gazillion/etc) were meaningless unless they decreased memory buss <> system buss latency. As we remember, huge decreases in internal module latencies and equally huge increases in ram "speed" were required to produce decreases in this latency metric . . . and it took equally huge decreases in this latency metric (2X) to produce only arguable increases in FS performance. Great fun for the computer hobbiest, but a high risk/limited reward proposition for the FS-er. I ran some FS tests with HD transfer rates of ~ 250MB/s then with at 100MB/s. There was no perceivable ingame result and load times were only marginally affected. Overall, my system "felt" more responsive with the 250MB/s transfer configuration. but this was sensed only because one tends to develop a very tactile sense about their own system. Again, the FS change was indecipherable. These test indicate that HD transfer rates at ~ 100MB/s are more than FS needs for any aspect ingame play. Faster transfer rates might help game/flight load times, but by only a few seconds, at best.
Sam,Did your test include flying over a rural area into a very dense urban area with a big add on airport? I guess this change will result in a lot of disk access reading a lot of scenery and texture files. No difference at all in such a scenario?

Share this post


Link to post
Share on other sites
From a more common sense point of view, one might consider that a hard drive's job is to transfer data to (and from) the system. That's the long of it . . . and the short of it. The rate at which the HD can transfer data to and from the system (ie data rate) will be the OnLy measurement parameter that will matter to a final performance result. For instance, the data must be accesses - before - it can be transferred to the system. Therefore, decreased access time (most certainly) has the potential to Increase data transfer rates, but as a stand-alone metric, it is meaningless. This is very similar to the ram speed discussion. Internal ram latency decreases only matter if they assist by Increasing the amount of data that can be transferred from/to the system. In that case, the controlling metric was memory buss <> system buss latency through the memory controller. Ram module internal timings (5-5-5-12/etc) and speeds (DDR12-50gazillion/etc) were meaningless unless they decreased memory buss <> system buss latency. As we remember, huge decreases in internal module latencies and equally huge increases in ram "speed" were required to produce decreases in this latency metric . . . and it took equally huge decreases in this latency metric (2X) to produce only arguable increases in FS performance. Great fun for the computer hobbiest, but a high risk/limited reward proposition for the FS-er. I ran some FS tests with HD transfer rates of ~ 250MB/s then with at 100MB/s. There was no perceivable ingame result and load times were only marginally affected. Overall, my system "felt" more responsive with the 250MB/s transfer configuration. but this was sensed only because one tends to develop a very tactile sense about their own system. Again, the FS change was indecipherable. These test indicate that HD transfer rates at ~ 100MB/s are more than FS needs for any aspect ingame play. Faster transfer rates might help game/flight load times, but by only a few seconds, at best.
Just like everything else.. you always come up with the "theoretical' argument and never, ever not ONCE have provided anything in this forum that makes a solid, defined positive difference to MSFS and what people really want out of itYou post the same bull over and over... you say you test a card or a drive but you do not have those new components placed in a system with a process/memory/storage subsystem that will actually take advantage of it.. then turn around and claim its all snake oilI have seen your 3-5 mile blurry images with little or missing autogenAt best you are guessing based on a very poor and not at all thought out approach. This is the mark of a layman and not anyone who is scientifically, technically or professionally oriented. When are you going to get it through your head that people here who have evolved past your "all you ever need" Q6600/DDR2 800/8800GT suggestions and who are now selecting the support components that will supply the base system at the rate it can process information are enjoying FSX more than you ever will?Never!Not until the day that you actually purchase a correctly spec'ed DDR3 system and clock it correctly. And even then, you would be too embarrassed to come into this forum and admit the expert from best-buy is wrong.So Sam.. your hit-n-run follow ups after me in threads are getting old. I doubt many are listening to this dribble anymore.To everyone reading this:FSX is not ALL about The processorThe memoryThe motherboardThe video cardThe hard driveThe OSThe setup including the clockIt’s about ALL of the aboveWhen you wish to purchase parts if you only run a Q6600/DDR2 800 88/9800GT you can use a 500GB large platter HDD however the WD Vrap WILL IN FACT eliminate access stutters and reduce CPU overhead which you may not know are being caused by storage system. The better the system the drive is attached to, including the SATA/SAS/SCSI interface, the better the result. You buy cheap, you get cheap and you get to sit there and wonder...was that my CPU/memory or my hard drive causing that?It’s a matter of defining what it is you are trying to accomplish with the least issues to deal with.

Share this post


Link to post
Share on other sites

hmmmmI think from a financial POV buying a Seagate 32mb cache HDD would be best price-perf ratio

Share this post


Link to post
Share on other sites
hmmmmI think from a financial POV buying a Seagate 32mb cache HDD would be best price-perf ratio
I understand the need to spend very carefully in today's economy. I also realize not everyone can afford these high-end parts. Unfortunately, going budget with FSX will not get anyone the performance that's available with the right hardware. Besides, that Seagate is still no match for the Vraptor for the sim. Probably closer to the first generation raptors and even then it would still probably lose in random access. My point is you get what you pay for as always.You want a purpose-built system for FSX. As close to a desktop supercomputer as you can get these days, to put it frankly. Sure, it'll run on far less but you'll get far less out of it. It needs horsepower and tuning. Cheap out on parts and you'll get a muddy, blurry, stuttery, less detailed simulator. I've been there-done that. And Intel didn't just plan to screw us simmers out of our money by forcing us to buy expensive parts. You can thank Microsoft for putting out FSX in the design state that it is in. Thanks to that we need hardware that can overcome the burden the software creates.-jk

Share this post


Link to post
Share on other sites

Anything beyond a modern 100MB/s HD will provide FS an entirely irrelevant performance advantage. Under any FS environmental circumstance, I found no difference with big airports/hi AG/addon airplanes/etc. Those higher transfer rate and lower latency (and Hiii dollar!) card-based raids n' raptors are a non-factor. Remember, it's all about moving data between the HD and the system (transfer rate). All those HD internal factors (Access times/Latencies/burst rate/etc) only assist that final result.

Share this post


Link to post
Share on other sites
Anything beyond a modern 100MB/s HD will provide FS an entirely irrelevant performance advantage. Under any FS environmental circumstance, I found no difference with big airports/hi AG/addon airplanes/etc. Those higher transfer rate and lower latency (and Hiii dollar!) card-based raids n' raptors are a non-factor. Remember, it's all about moving data between the HD and the system (transfer rate). All those HD internal factors (Access times/Latencies/burst rate/etc) only assist that final result.
Blah, Blah, Blah...Sam you don't know what's about what here. You're blowing hot air, nothing more. You don't have the parts OR the credentials to give advice here. And most of what you say is so wrong it should be criminal."I will not eat green eggs & ham...I will not eat them, Sam I am."When are you going to back up your arguments the way Nick has been challenging you to? -jk

Share this post


Link to post
Share on other sites
From a more common sense point of view, one might consider that a hard drive's job is to transfer data to (and from) the system. That's the long of it . . . and the short of it. The rate at which the HD can transfer data to and from the system (ie data rate) will be the OnLy measurement parameter that will matter to a final performance result. For instance, the data must be accesses - before - it can be transferred to the system. Therefore, decreased access time (most certainly) has the potential to Increase data transfer rates, but as a stand-alone metric, it is meaningless. This is very similar to the ram speed discussion. Internal ram latency decreases only matter if they assist by Increasing the amount of data that can be transferred from/to the system. In that case, the controlling metric was memory buss <> system buss latency through the memory controller. Ram module internal timings (5-5-5-12/etc) and speeds (DDR12-50gazillion/etc) were meaningless unless they decreased memory buss <> system buss latency. As we remember, huge decreases in internal module latencies and equally huge increases in ram "speed" were required to produce decreases in this latency metric . . . and it took equally huge decreases in this latency metric (2X) to produce only arguable increases in FS performance. Great fun for the computer hobbiest, but a high risk/limited reward proposition for the FS-er. I ran some FS tests with HD transfer rates of ~ 250MB/s then with at 100MB/s. There was no perceivable ingame result and load times were only marginally affected. Overall, my system "felt" more responsive with the 250MB/s transfer configuration. but this was sensed only because one tends to develop a very tactile sense about their own system. Again, the FS change was indecipherable. These test indicate that HD transfer rates at ~ 100MB/s are more than FS needs for any aspect ingame play. Faster transfer rates might help game/flight load times, but by only a few seconds, at best.
Sam,Did your test include flying over a rural area into a very dense urban area with a big add on airport? I guess this change will result in a lot of disk access reading a lot of scenery and texture files. No difference at all in such a scenario?What about the ohter hardware when testing? What CPU, mobo, RAM, graphics card did you use? OC:ed?

Share this post


Link to post
Share on other sites
Scroll down to the end of the threadhttp://www.simforums.com/forums/forum_posts.asp?TID=29041chart says it allSAS/SCSI are better suited for multiuser accessthe best drive is the WD Vrap
Yes, I see that. Even so Nick, you make this comment in your article: "Velociraptors in RAID0 (256K STRIPE) on the right card are very close to SCSI performance"I have a 15K.5 SCSI right now already on a U320 controller (it's running 180 in single channel mode I think). The comment above suggests my SCSI should be ample for FSX, or am I misinterpreting something? I had been running FSX in XP on that drive, however I haven't moved my Vista 64 over to FSX on that SCSI, and was wondering if it would be worth it. Sounds like it may be, though again, it's isn't too bad on my sluggo SATA II drive.Thanks in advance . . .Noel

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.