Move gaming folder to external drive for better performance? See data:

[Obersturmfuhrer]

Hello,

I would say I'm an average gamer running a pent. core 2 duo 2.26 Ghz, 4MB RAM, 1GB Video memory from nvidia GT 130M GPU, with 500 GB 5,400 RPM hard drive. Using laptop as a "desktop replacement" and Running win7 x64, have my OS on C:\ drive along with (Currently) Steam folder and Games for Windows - LIVE games.

I also have a 1TB eSATA external hard drive that, according to HD Tune Pro app, read times are faster than in the internal drive.

The disparity isn't much, however, as you can see by the pics of result screen i've provided.

Here is the 1TB external Drive data:

Here is the 500GB internal data:

As shown, the ext. drive is overall around 10/MB/s faster than the internal. Unfortunately, I do not know how to interpret this information. Would this 10 MB/s yield a considerable performance increase in gaming? Even in the slightest manner? Would the results be even noticeable? I would have to go to a fair amount of trouble moving the data from one drive to the other so I'd like some concrete information on this topic before I go to the trouble. If load times will decrease by a billionth of a second, I'd prefer to leave the games where they are.

One last point to consider would be the following, I understand also that whichever drive hosts ones OS will perform better with less overall data/information on it. Seeing in my case that my internal C: drive (unpartitioned) plays host to my win7, this could be another advantage of transferring my games to the external drive. In total, I have between 70-100GB of game data on my drive, which will also get bigger in time.

I have the drive connected via the eSATA port on my laptop, in order to achieve faster transfer rates than USB 2.0 obviously.

Thanks for reading, if anyone could help me out here I would greatly appreciate it!

OBER.

 

[buwish]

I think that you would see your performance fall, regardless of how good the external drive is. Remember, the internal drive is going to perform better than an external drive, especially one connected via USB 2.0. USB 2.0 can only handle up to 480MB/s (in theory), while the internal SATA controllers can deal with data at 3GB/s. Now, if you were talking about USB 3.0, it might be a bit different.

 

[homiezheadsup]

Being that you claim your external is eSata I'm going to say you will see a improvement on loading times by loading game data on it. But I doubt you will see any frame rate improvements for that is 99% cpu/gpu/ram dependent.
On the other hand:
I do not like using external drives to store software info because it adds another point of failure. If you're half way through your game and all of a sudden you accidentally trip on the wire and it unplugs the power or esata cable, then your computer will probably freeze, and it is highly likely you can cause damage to your eHD.

 

[Obersturmfuhrer]

I think that you would see your performance fall, regardless of how good the external drive is. Remember, the internal drive is going to perform better than an external drive, especially one connected via USB 2.0. USB 2.0 can only handle up to 480MB/s (in theory), while the internal SATA controllers can deal with data at 3GB/s. Now, if you were talking about USB 3.0, it might be a bit different.

I stated I wasn't using USB 2.0, it is connected via eSATA.

As for the other reply, yes I realize there'd little or no framerate improvement as that is entirely dependent upon CPU/GPU like you say. However, I am speaking in just load times, data access times, etc... Thus would improve overall game performance. As for tripping over a power cord or some random accident, that's highly unlikely given the way my particular office/desk setup is. I see what you are saying, thou, perhaps if the power went out or something like that.

What about the issue of keeping little data possible on the drive that the OS is located on? I read that this increases overall boot times, computing performance as a whole.

Bottom line is, should I do it or not? Would you? (referring to anyone reading this)

I mean, the numbers don't lie, right? The external eSATA has faster read times than the internal drive, but I Suppose the main question is, is 10MB/sec a noticeable difference, even in the slightest, with regards to performance?

 

[homiezheadsup]

Simple reply is yes, you should do it.
You will see faster loading times for having 1 hd for os software load while the other is preping the game content. Along with doing multi tasking for you can have the eHD load the game while chatting on AIM with your OS harddrive.

I would not do it, for I cannot ever setup a eHD to be full proof.

 

[bakareshi]

As long as you take care of your eHDD you should have no problems here. eSATA transfer times are on par with SATA II and games run similar on both. Note that other factors come into play with platter drives such as latency and seek times.

You will see performance improvements with your internal drive after a mass exodus of your game files. This will free up real estate on the outer sectors of the hard drive platters where the best performance is found. After defragmenting with a decent tool like diskeeper, don't bother with the defragment utility bundled with Windows as it is complete trash, you can consolidate your files, most importantly the page file, to those precious outer sectors of the HDD platters to get the most performance from your HDDs.

Once you accomplish all of that you will notice a nice performance increase. You will be on your way to becoming a hard drive performance junky. Then you will seek a decent 7200 RPM drive to replace that lump that your laptop currently boots from. Or, you can skip that and go straight to the logical endpoint and install one of the newer SSDs as your boot/application drive. I press the power button on my Windows 7 Ultimate machine and in less than 35 seconds I am in Windows and surfing the web. That includes going through the login screen.

 

[MRFS]

There is no difference between SATA and eSATA, as far as
raw bandwidth is concerned. The eSATA standard supports
a slightly higher voltage, so that external SATA cables can
be longer.

Here's a much better solution: "short-stroke" your faster HDD.
In this case, it's your 1TB.

You can shrink your C: partition using the PartitionWizard freeware:

http://www.partitionwizard.com/


Then, all I/O to and from your C: partition will fall to the left of
a vertical line somewhere between 0 and 10% on your HDTune graphs.

Here's a really great study showing the effects of various recording densities:

The HDDs with PMR (perpendicular magnetic recording) fall slower,
because their tracks are much closer together.

Raw transmission speed i.e. "buffer to/from disk", is directly proportional
to track circumference, because all modern HDDs try to maintain
the same or similar recording density from outermost to innermost tracks:
hence, the "fall" that you observe in all HDTune graphs like this:

http://img3.imageshack.us/i/1tbdrivebenchmark.png/


MRFS

 

[Obersturmfuhrer]

There is no difference between SATA and eSATA, as far as
raw bandwidth is concerned. The eSATA standard supports
a slightly higher voltage, so that external SATA cables can
be longer.

Here's a much better solution: "short-stroke" your faster HDD.
In this case, it's your 1TB.

You can shrink your C: partition using the PartitionWizard freeware:

http://www.partitionwizard.com/


Then, all I/O to and from your C: partition will fall to the left of
a vertical line somewhere between 0 and 10% on your HDTune graphs.

Here's a really great study showing the effects of various recording densities:

http://www.supremelaw.org/systems/io.tests/platter.transfer.crossover.graphs.2.png


The HDDs with PMR (perpendicular magnetic recording) fall slower,
because their tracks are much closer together.

Raw transmission speed i.e. "buffer to/from disk", is directly proportional
to track circumference, because all modern HDDs try to maintain
the same or similar recording density from outermost to innermost tracks:
hence, the "fall" that you observe in all HDTune graphs like this:

http://img3.imageshack.us/i/1tbdrivebenchmark.png/


MRFS

Shrink my C:? I always thought generally it was better to keep a drive partition-free for best performance. This is interesting data you've proposed here, however, I think we're all focusing on the wrong issue here. It's not so much the comparison between SATA and eSATA transfer speeds; it's the raw read/write speed that is truly significant in my case. And clearly, my external drive is faster than my internal, in that regard. Whether the data is being processed faster by the MB/CPU I suppose remains to be seen.

Anyway, I've decided to proceed with the data transfer. Wish me luck, I hope everything goes smooth. As a side note, Steam does NOT recommend storing game data on an external hard drive, which can be seen here . They site, ironically, due to "performance" issues. I presume these "performance" issues may have to do with more practical than technical reasons, such as power-outages, external hard drive failure, etc., not because it will hinder gaming performance.

Thanks for the replies I appreciate all your input!

 

[bakareshi]

What MRFS is telling you to do, shrink your C:, will ensure that your data stays on the outer regions of the platters. There is nothing wrong with resizing your C drive. Considering the Caviar Black goes from 160 MB/s on the outer sectors to less than 80 MB/s on the inner sectors, there can be a huge performance loss from outer to inner regions on HDD platters. Your 1TB drive will be considerably slower reading from the inner sectors as well. We aren't trying to tell you that one drive is better than the other, just that your read speed will be higher if your info is on the outside of the disk platters and lower on the inside.

Your computer knows this and writes to the outer regions first, which is why performance decreases as your drive fills up and writes to the inner portion of the platters.

Imagine you put one-inch rings on a car tire from the outside edge of the sidewall to the inside edge. Then write messages in those stripes using the same size letters for all of them. The outermost ring will contain the most information and the amount of information on each ring will decrease until you reach the innermost ring which contains the least amount of information. Now imagine you can only read the messages back from one fixed point as the tire turns at 1RPM. Reading the outermost message will provide more information in 1RPM than reading any of the other rings. Keep in mind that it still takes you 1 minute to read a message regardless of which ring your read it from, because that is the speed the tire is turning.

HDDs work the same way. One rotation takes the same amount of time, regardless of what area of the platter is being read. The outer regions hold the most information, therefore one rotation from the outside area of a platter will yield the most information in a set amount of time.

Hence if you load a game located on the outer region of a platter and it takes 10 seconds you can expect the same game to take as much as 3 to 5 times longer if it is located on the inner region of a platter.

 

[Obersturmfuhrer]

Very interesting, I never really new or even really considered how a HDD actually stored data and processed it. I knew it was circular obviously but I didn't know that data was stored in continuous stripes along "rings" of the diameter of the platter. I saw the inner workings of a HDD once on some youtube video and the reader arm (not sure what it's technically referred to as) was rapily going to and fro the outermost to the inner most areas of the HDD. Seeing that, I figured that data was simply randomly stored and/or may have gotten arranged in such a way after copying/deleting/moving various files.

In regards to establishing a new partition or shrinking the C drive, what point is optimal to do this at? Right after a fresh win7/OS install? Is there a specific way to "tell" a partitioning program isolate various areas of the platter? How much control does one have over where a paritition is established and/or created?

Nevertheless, great info baki & MRFS, I eat stuff like that up!!

 

[bakareshi]

Like I said in an early post, your HDD is automatically designed to use the outermost regions first to optimize for performance. On a fresh install if you shrink the drive it will automatically partition to the outer regions first. With a decent disk care program like diskeeper you will have more control over where files are stored on your disk. Your assumption that files get spread out over the disk is correct. This comes from normal use as you add and remove files and is the reason a defragment will speed up your drive.

 

[hang-the-9]

Your internal laptop drive is 5400 rpm, the external one is probably 7200 if it has an esata connections. There is the difference in speeds. If you are worried about system preformance, replace your 5400 with a faster laptop drive. I do think they make 7200 rpm laptop drives. Or even an SSD, that will speed up your whole system.

 

[Obersturmfuhrer]

Your internal laptop drive is 5400 rpm, the external one is probably 7200 if it has an esata connections. There is the difference in speeds. If you are worried about system preformance, replace your 5400 with a faster laptop drive. I do think they make 7200 rpm laptop drives. Or even an SSD, that will speed up your whole system.

No, the external drive is 5,400 as well. Why the increase in speed read spead according to the graphs? I don't know why; like I said they are both indeed 5,400 RPM. Perhaps the external drive simply runs more efficiently. If it was 7200 RPM I am sure it would be MUCH faster than it is.

I realize that a SSD would be faster, however, the data and videos I've seen displaying their capabilities and differences between them and regular drives does not justify nor is worth the extra expense, particularly for the comparably limited amount of data they hold. (128GB compared to 500 or 100GB for similar cost range) To be completely honest, for the SSD's I've seen, the speeds and performance haven't exactly "knocked my socks off". I believe, when combined with other superior computer components, the system as a whole would function impressively. This is the manner in which I believe SSD's should be utilized in.

 

[hang-the-9]

If both drives are the same RPM, the difference could just be in the different models. Unless you are comparing the exact same drive using different connections, you can't go by the benchmarks.

 

[Obersturmfuhrer]

If both drives are the same RPM, the difference could just be in the different models. Unless you are comparing the exact same drive using different connections, you can't go by the benchmarks.

So you're saying that all 5,400 RPM drives should have the same read/write times? I Dont think so, I haven't researched this but I am sure there are other factors that go into overall performance other than just RPM speed.

 

[hang-the-9]

So you're saying that all 5,400 RPM drives should have the same read/write times? I Dont think so, I haven't researched this but I am sure there are other factors that go into overall performance other than just RPM speed.

Nope, I'm saying that if the RPMs are the same, it's other factors with the drive that cause the differences. Cache sizes, firmware, platter density, etc... RPM was really just the most obvious one.