AMD Phenom II X4 965 125W Black Edition Overclocked Benchmarking

After overclocking, testing and benchmarking for days, on AMD Phenom II X4 965 125W Black Edition CPU, here we get to publish the results. Just keep in mind that we're not comparing video graphics adapters but CPUs so the differences between scores, vary only based on how much the benchmark is CPU dependent. In some cases they are minor in others major. 


System Specifications :


ASUS M3N78-VM
ATi Sapphire Radeon HD4830 512MB Overclocked @ 740/1291 MHz
2x2GB Ram DDR2 Dual Channel Overclocked @ 1024MHz
(Add Air Condition to the cooling... It makes a big difference. ;-) )


Here we go : 


The CPUs in comparison are the following :


AMD Phenom II 965 BE X4 - Overclocked @ 4.363MHz
AMD Phenom II 965 BE X4 - Default Clocks @ 3400MHz
AMD Athlon 5200+ X2 - Overclocked @ 3200MHz


7-zip Benchmark
Measured in MIPS (Million Instructions Per Second)

As we can see in the simple 7z benchmark, the overclocked 965 processor is clearly the winner no matter the number threads in use by the system. Note that 5200+ is a dual core CPU and that's why there is only a slight difference between its score achieved at 2 and 4 threads. But also note the superiority of the overclocked 965 over the default clocked 965 on all 3 cases.

Cyberlink PowerDirector 9.0 - Video Converting

Measured in seconds - Lower is Better

In this test, we simply tried upscaling while changing the encoding type of a high definition video file from 720p .mov format to a 1080p .mp4.

Playback duration : 4 mins 5 sec

Details : 

MOV File Details: (File to be converted)

Video Resolution : 1280x720 @ 30 FPS

Data Rate : 15042kbps

File Size : 443MB

MP4 File Details: (Converted file)

Video Resolution : 1920x1080 @ 25 FPS

Data Rate : 11521kbps

File Size : 341MB

Time taken on 5200+ @ 3200MHz : 6 mins 14 sec

Time taken on 965 @ 3400MHz : 4 mins 34 sec

Time taken on 965 @ 4363MHz : 3 mins 59 sec (Even less than the playback time!!!)

As we can see here the default 965 took only 73% of the time taken by the 5200+. More over the overclocked 965 took only 63% of the 5200+ time. Now keep in mind that this test was done on a small duration video, only 4 minutes and 5 seconds.. in case you would like to do the same test on a normal movie (2 hours duration) the time differences would be much bigger. If you do the maths you'd find out that a 2 hour video using the same data rates would take 3 hours and 2 minutes on the 5200+, while on the overclocked 965 it would only take 1 hour and 54 minutes. Yep.. i bet you get the picture. :-)

Benchmarks

In this section we tested the three CPUs using some commonly accepted benchmarking tools. And when the word "Benchmark" enters our ears, there is only one thing that comes up in our brain.. 3D Mark. For more than 10 years now it has been the basic comparison tool for for most of the pc enthusiasts all over the world. So our review would be incomplete without adding them to the list. We tested 3D Mark 2005, 2006 and Vantage using the three CPUs in our testing list. Let's check the results out..

Keep in mind that the most important bar in all three following charts is the CPU Score one. At least for our review as our comparison is between CPUs and not VGAs.

3D Mark 2005

Measured in 3DMarks - Higher is Better

Check this out.. The 965 overclocked edition is much more than 2 times faster than the 5200+ and 28% faster than the default clocked edition. As we can see the difference between the 3 processors is really high according to 3D Mark's tests. Let's see more of them..

3D Mark 2006

Measured in 3DMarks - Higher is Better

3D Mark 2006 was always known to be really stressing our CPUs and it's still very hard to achieve high scores on its CPU tests. However once again we see that the overclocked edition is almost 3 times faster than the 5200+ and 27% faster than the non-overclocked edition. It's also worth mentioning that the final score shows a 4794 points increase over the old 5200+.

3D Mark Vantage

Measured in 3DMarks - Higher is Better

Once again the overclocked edition of the 965 is 3 times faster than the 5200+ and 26% faster than its default clocks edition. So, let's sum it up..

As we can see the performance difference between the 3 processors is massive when it comes to benchmarking. The 2 extra cores that the 965 has, surely help a lot, but we can also notice that overclocking it helps performance even more. The difference is still huge when we compare the overclocked edition of 965 with the non-overclocked one in these three benchmarks. However these scores don't necessarily define how the system would perform against against some "real life" gaming, but they are still good in order to compare older CPUs with the latest ones. So, we couldn't help it but do some "real" gaming with some of the latest games in the market and share the results with you. Read on..

Now let's talk about games...

We did a lot of benchmarking on some of the latest games on the market, measuring the average, minimum and maximum frames per second the video graphics adapter was drawing at each time, on different game locations and graphic details. After collecting the results we got the averages of each case and then we created the charts so you can clearly view the results.

All test were done using a screen resolution of 1920x1080 using Directx Version 10.1.

All temperatures were ~24 Celsius degrees at all times using air-condition cooling.

All games were set at the maximum detail level possible changing only the Anti-Aliasing levels through each test.

Crysis 2

Crysis 2 has been out for a while, and it makes use of most of the current technologies in the graphics section, so it would be a nice way to benchmark our processors. So here we go. We used the general options (Gamer, Advanced, Hardcore) in the Video Settings of the game just like what most pc users would do. (There is a way to unlock the advanced options but only a few gamers around would try it, so we try to keep the tests as close to real life gaming as possible).

Crysis 2 Gamer

Measured in Frames Per Second - Higher is Better

Crysis 2 Advanced

Measured in Frames Per Second - Higher is Better

Crysis 2 Hardcore

Measured in Frames Per Second - Higher is Better

As we can see in these three tests the overclocked 965 gets in the first place, gaining up to 9 FPS depending on the graphics quality. Surely a better graphics card would perform much better but still we can easily notice the difference between the 3 processors.

Red Faction Armageddon

One of the latest and very demanding game we decided to add to the benchmarks is Red Faction Armageddon. It is an open world 3rd Person Shooter-Future RPG with amazing physics allowing you to destroy and re-create almost everything in the map. We tested the processors in three different cases. 

Red Faction Armageddon - No Anti-Aliasing

Measured in Frames Per Second - Higher is Better

Red Faction Armageddon - 4 x Anti-Aliasing

Measured in Frames Per Second - Higher is Better

Red Faction Armageddon - 8 x Anti-Aliasing

Measured in Frames Per Second - Higher is Better

As you might have noticed the specific system barely makes it to keep the average of the 30 FPS playable limit, and that happens only with the overclocked 965, and only with anti-aliasing disabled. Once again you can notice the difference in FPS between all three processors. You should also keep in mind that you need a newer graphics card in order to enjoy the experience of this game at maxed detail.

Test Drive Unlimited 2

That's also a gaming choice we'd like to include mainly because in racing games it's all about FPS, if you know what we mean. And this game would be a fine benchmark tool because of the dynamically designed in-game open world and fast pace gameplay. It surely needs a beast of a gaming rig using one of the latest high end graphic cards in order to run properly with no lagging. You'll see what I mean once you check out the tests below.

Test Drive Unlimited 2 - No Anti-aliasing

Measured in Frames Per Second - Higher is Better

We see that the system performs overly well but still in some cases at high speed driving we see a performance drop of about 6 frames. As we compare the 3 CPUs, the overclocked 965 performs pretty well staying above 30 FPS for most of the time. It's non overclocked edition does also keep up pretty well only it drops more when the load is high. Now as for the 5200+ we could say that this game is almost unplayable at full detail level. Now let's turn that anti-aliasing switch on..

Test Drive Unlimited 2 - 4x Anti-aliasing

Measured in Frames Per Second - Higher is Better

No big changes, 4x Anti-aliasing seems easy. We notice an overall small drop of about 4 FPS on the 965 while the 5200+ stays almost untouched. However the difference between the overclocked and non-overclocked edition of the 965 becomes even smaller. Remember that we are running the game at maxed details except V-Sync. Check out the 8x Anti-aliasing test below..

Test Drive Unlimited 2 - 8x Anti-aliasing

Measured in Frames Per Second - Higher is Better

Well the differences are now somewhat tragic. There is a possibility that something went completely wrong with the game when trying on the 5200+. So let's keep that out of the comparison. Now as for the 965 we see another ~4 FPS drop when running at 8x Anti-Aliasing. And as we can see the ATi 4830 is still fighting.. though the game becomes unplayable. Also keep in mind that it's no ordinary 4830.. it's superclocked at 740/1290 MHz from 575/900 and also it's working under 24 degrees so the performance difference is massive..

Mass Effect 2

Now this was a choice mostly for fun than comparison. This game is not really demanding but we added it as a reference because many gamers around the world find it interesting and they have to know about how this system would behave while running it. Keep in mind that Mass Effect 2 uses a frame-rate limit at 60 FPS. We decided to leave it as is because most gamers wouldn't bother removing that. It's actually pretty smooth playing at 60 FPS. So let's see how it performs.

Mass Effect 2 - Full Detail Level (There's no in-game anti-aliasing option! You could enable it manually on AMD Vision Engine Control Center if you'd like to..)

Measured in Frames Per Second - Higher is Better

The game is smoothly playable on both processors and the non-overclocked edition of the 965 seems to be even more stable. However we can see a fine difference of ~8 FPS between the 5200+ and 965.. Anyway, as we mentioned, this test is just to let you know that with a decent graphics card you can play it at maxed details.

Conclusion

We bet that concludes all the tests. We guess the upgrade from the 5200+ was amazing including the price-performance ratio. The only thing that's stopping us now from the newest games is a better graphics card. Also we might use these results in order to post the difference levels after I upgrade. The AMD Phenom II 965 Black Edition is an amazing CPU even non-overclocked but for enthusiasts.. it's hell of an overclocker. Highly recommended!

Thanks for reading!

Solid Snake

AMD Phenom II X4 965 125W Black Edition Extreme Overclocking

Hello everyone, this is Solid Snake, and we've been busy for a long time preparing this new review about my latest CPU Upgrade. we're talking about my brand new AMD Phenom II X4 965 125W Black Edition. This beast was definitely worth its money and proved to be an extreme overclocker.. and by saying it, we really mean it. 

First of all let's take a look at it's manufacturer specifications :


CPU Type : AMD Phenom II X4 Quad-Core
Model : Phenom II 965
Code Name : Deneb
Core Speed : 4 x 3400 MHz
Revision : C3
Manufacturer Number : HDZ965FBGMBOX
TDP : 125W
Vcore : 0.825 - 1.4
Maximum Operating Temperature : 55-62
System Bus : 4000HT/S
L1 Cache : 512KB (4 x 128KB)
L2 Cache : 2048 (4 x 512KB)
L3 Cache : 6144KB
Socket : AM3
Processor Achitecture : 45nm SOI
Operating Mode : 32/64 bit


Plus don't forget that this is a Black Edition CPU which means it has an unlocked multiplier which far extends the oveclockability possibilities. Now let's get down to the process.


As this CPU was an upgrade to my previous one we had to compare these two in order to check the performance difference and upgrade value. So before getting my beloved AMD Athlon 5200+ , which has been serving me well for 3 years now, off my mainboard, we decided to overclock it as far as possible for one last time using extreme conditions, make a lot of benchmarks and then do the same to my new AMD Phenom II 965 one..


For extreme overclocking we've used Air Condition Cooling which gave us perfect temperature values down to 17-24 degrees during the whole process (I almost got a cold that day :-) ).


As for my system specifications you can check our previous post here :
My System Specifications.


As for the tools We've used, the following list includes them all : 


Frequency - Temperature Monitoring :


CPUID CPU-Z 1.57.1
CoreTemp 0.99.8
GPUTool 0.0.0


Compression - Conversion Tools :


7-zip
Cyberlink PowerDirector 9.0


Benchmarking Tools :


3D-Mark 2005
3D-Mark 2006
3D-Mark Vantage


Games :


Crysis 2
Red Faction Armageddon
Test Drive Unlimited 2
Mass Effect 2


using Fraps V.3.2.3 Build 11796 to collect the FPS values.


Also lets state that we used Windows 7 Ultimate x64 and that VGA was upgraded to its latest drivers:
AMD VISION Engine Control Center V.11.6


So after days of benchmarking my old AMD Athlon 5200+ it was time to switch to the new AMD Phenom II 965. So we downloaded and flashed my mainboard to the latest BIOS available, entered safe mode, uninstalled the previous processor from the device manager and turned off the whole system. We got the Xigmatek HDT-S963 cooler out, then the old CPU. Cleaned the cooler surface, inserted the AMD Phenom II 965 in the socket, applied some fresh XILENCE ZUB-XPTP .X5 X5 thermal grease, and installed the cooler back in. After everything was set up correctly, we powered on the system and entered bios to check if the readings were correct. Then we entered Windows 7 and re-did the benchmarks using the default settings with no overclock applied.


Then once more we got in to BIOS in order to check the maximum overclock levels of the specific processor. After hours of playing around with the frequencies and multipliers we managed to reach the limit of 4491.76 MHz and validated it with CPU-Z. Check the validation here : CPU-Z Validation. This was terrifying at first, keeping in mind that the default frequency of the processor was 3400 MHz. This means almost 1100MHz or 32% overclock on the default frequency. Now as for the stability testing the system performed nicely but was not totally stable under huge load. So we decided to keep the frequency of 4363 MHz as the stable frequency for benchmarking. Here are some more details on the BIOS setup as shown on CPU-Z: 


FSB : 256
Core Multiplier : x 17.0
Core Frequency : 4363.4 MHz (Effective -> x 4)
Core Voltage : 1.456 V (+150mV on default)
Ram Frequency : 1024 MHz (overclocked from 800MHz)
HT Link : 1283.4 MHz (we let that one down in order to keep the system fully stable)
CPU Temperatures : 24 - 41 degrees (A/C cooling + Xigmatek HDT-S963)


So this is the time for the final comparison :


Benchmarking!

ATi Sapphire HD4830 512MB GDDR3 Overclocking And Benchmarking

Hello everyone it's been a long time since I made my last post. Hopefully my next posts will come more often.


I finally decided to post my overclock results for my main graphics card, ATi Sapphire HD4830 512MB GDDR3. I've been using it for more than 3 years now and I still don't feel the need to purchase a new card cause I still play most games at maxed details with no problems. It's really a beast of a Graphics Card.. Let's check the details:


Part Specifications:


Chipset: Ati Radeon HD 4830
Memory Size: 512MB
Memory Type: GDDR3
Memory Interface: 256-bit
Core Clock: 575 MHz
Memory Clock: 2x 900 MHz
Connection Type: PCIEx16
Maximum Resolution: 2560x1600
DirectX Support: 10.1
Shader Model Support: 4.1
Output Ports: 2x DVI, 1x Display Port


My System Specifications:


CPU: AMD Athlon X2 5200+ Brisbane
(Overclocked from 2.7GHz to 3.2GHz)


RAM: 4GB (2x2GB) DDR2
(Overclocked from 667MHz to 948MHz)


For details on other parts please check my other post : My System Specifications


OS: Windows 7 Ultimate 64-bit


Driver Version
CCC2 Beta - 10.12


Overclocking Utility
GPU Tool CCT1 V.0.0.0


Applications:
3D Mark 2006 V.1.1.0
(using the SystemInfo Hotfix Feb08)
3D Mark Vantage


Overclocking Process:


The whole overclocking operation took place on my main desktop pc which is stable on high overclocking conditions. As you can see my CPU and RAM are highly overclocked, and to achive this I've used air cooling solutions such as a Xigmatek 963 CPU Cooler and Corsair Dominator Airflow and new heatsinks on Ram modules. Voltage was also raised properly on both of those, and still temperature stays below 45 degrees under high load conditions. The ATi Sapphire HD4830 512MB comes with a fine stock cooler. It's a huge round heatsink with a fan. I did not need to replace it as it does an excellent job keeping the VGA card cool. To be more specific the card operates with no load (on windows) at 40-43 degrees and at high load (3D Mark-Games) at 55-65 degrees. Fan speed was set to 100% during the overclocking and got a bit noisy but I could live with that.


So in order to find out the max core levels that the HD4830 can reach I used GPU Tool's "Find Max" functions. The core started getting raised a few MHz at each time and soon it got above 700MHz.. Then it kept raising and passing heatup phases until it reached 755 MHz which caused a few artifacts on the screen. So the maximum overclock for the core without artifacts was around 750MHz which means 225MHz overclock or ~40% raise on the default frequency (pretty nice ain't it?). Then the memory followed the same process and the limit the overclock reached before failing was 1255MHz which means 355MHz overclock or also ~40% raise on the default memory clock. Now keep in mind that the effective clock on the memory is double that value.. As we can see the result is that the ATi Sapphire HD4830 512MB can be overclocked using the stock cooler at about 40% higher clocks. In order to avoid failures I decided to set as a max value for the core clock the 740MHz and for the memory clock the 1240MHz. Now how about some benchmarking?


3D Mark 2006
Default Settings


(Default Clocks)
Core Clock : 575MHz
Memory Clock : 900MHz (Effective 1800MHz)


3DMark Score : 10072
SM2.0 Score : 4325
HDR/SM3.0 Score : 4960
CPU Score : 2303


(High Overclock)
Core Clock : 700MHz
Memory Clock : 1200MHz (Effective 2400MHz)


3DMark Score : 10968
SM2.0 Score : 4843
HDR/SM3.0 Score : 5581
CPU Score : 2313


(Maximum Overclock)
Core Clock : 740MHz
Memory Clock : 1240MHz (Effective 2480MHz)


3DMark Score : 11367
SM2.0 Score : 5040
HDR/SM3.0 Score : 5835
CPU Score : 2358


3DMark Vantage
Default Settings


(Default Clocks)
Core Clock : 575MHz
Memory Clock : 900MHz (Effective 1800MHz)


3DMark Score : P5468
GPU Score : 5767
CPU Score : 4734


Details:
GPU Test 1 : 16.46 FPS
GPU Test 2 : 17.34 FPS
CPU Test 1 : 594.35 Plans/s
CPU Test 2 : 7.80 Steps/s
Feat. Test 1 : 481.04 GTexels/s
Feat. Test 2 : 3.09 GPixels/s
Feat. Test 3 : 15.99 FPS
Feat. Test 4 : 16.51 FPS
Feat. Test 5 : 27.99 FPS
Feat. Test 6 : 32.59 FPS


(High Overclock)
Core Clock : 700MHz
Memory Clock : 1200MHz (Effective 2400MHz)


3DMark Score : P6460
GPU Score : 7306
CPU Score : 4795


Details:
GPU Test 1 : 20.80 FPS
GPU Test 2 : 22.02 FPS
CPU Test 1 : 597.19 Plans/s
CPU Test 2 : 8.02 Steps/s
Feat. Test 1 : 584.85 GTexels/s
Feat. Test 2 : 4.15 GPixels/s
Feat. Test 3 : 19.44 FPS
Feat. Test 4 : 20.20 FPS
Feat. Test 5 : 32.13 FPS
Feat. Test 6 : 39.63 FPS


(Maximum Overclock)
Core Clock : 740MHz
Memory Clock : 1240MHz (Effective 2480MHz)


3DMark Score : P6651
GPU Score : 7637
CPU Score : 4796


Details:
GPU Test 1 : 21.99 FPS
GPU Test 2 : 22.76 FPS
CPU Test 1 : 595.90 Plans/s
CPU Test 2 : 8.06 Steps/s
Feat. Test 1 : 618.77 GTexels/s
Feat. Test 2 : 4.30 GPixels/s
Feat. Test 3 : 20.64 FPS
Feat. Test 4 : 21.26 FPS
Feat. Test 5 : 36.71 FPS
Feat. Test 6 : 41.88 FPS


Conclusion:
The performance difference between stock and overclocked clocks is highly noticable. Remember this is only a benchmark comparison. When it comes to everyday gaming the performance gain is even higher.


As you can see the ATi Sapphire HD4830 512MB is still a high end card plus after considering that it overclocks around 40% higher than its default clock values, using only its stock cooler the overclock makes it last even longer without the need of an upgrade. It performs well even when playing the today's market games at maxed details.


Some of the games I can run at maxed details right are Mass Effect 2, Alien VS Predators, Colin McRae DiRT 2, Batman Arkham Asylum, Tom Clansy's Splinter Cell Double Agent, James Cameron's Avatar, Mount & Blade Warband, Prototype, Fallout 3, Wolfenstein plus many many more.. So I guess I can stick to my ATi Sapphire HD4830 512MB for a while longer before upgrading.


That's all! I hope you enjoyed reading my article. Come back soon for more overclocking & benchmarking content!

What is your USB Flash Drive's Speed? Toshiba 8GB USB Flash Drive Testing

Hey everyone, a few days ago I purchased a new flash drive just like this one above in the picture, a Toshiba 8GB USB Flash Drive, and while having some free time, I decided to try out some custom, real life benchmarking with it, in order to get some facts about the real read/write speeds that it can accomplish.

Everyone nowadays owns a USB flash drive so it would be cool to know if we can modify one to get faster or something like that. First of all, let's take a look at the manufacturers specifications :

Drive Size : 8 GB

File System : FAT32 (anything above 2 GB on flash drives support only FAT32)

Default Cluster Size : 4 KB

Connectivity : USB 2.0/1.1

Read Spead : up to 18 MB/s

Write Spead : up to 4,5 MB/s

The only modification option we have on flash drives is to make small adjustments to its file system settings (cluster size) while formating from the computer management application (right click My Computer -> Manage -> Disk Management) and in my case the only adjustments I was able to make, were to format the USB flash drive using FAT32 picking a different cluster size each time, choosing from 2 KB up to 64 KB.

Afterwards I ran a lot of tests (took me about half a day to complete them all) using a stopwatch in order to check how much time does it take normally for us to copy some files in there, from our main hard disk drive. Note that every time we copy a file in the drive, the operating system splits each file in such small parts according to its cluster size setting while copying them in it. This means that in case the files we copy are smaller than the cluster size, they won't have to be fragmented, but in the case that they are bigger or much bigger they will have to be cut down to pieces of the cluster's size in order to be copied. We should also keep in mind that in case the cluster size is bigger than the files we copy, each file takes more disk space on the drive.

About the tests, first of all, I used three different types of files and all of the available cluster size settings running most of the test especially the important ones at least twice.. File size means the file's size according to the data, disk file size taken is usually more than the file size.. The system specifications are not high end as most people around don't have high end PCs plus copy pasting doesn't get affected much affected from hardware changes. It's mostly related to the drive's manufacturer's specifications. But just for the record it is a Pentium 4 2,66, 1 GB ram, running Windows XP. Also, I should state that I disabled all other software from running while benchmarking.. even the antivirus so that we won't get any delays with checks etc. Here are the test types..:

Test #1 :

File Type : One ISO file (.iso)

File Size : 1.20 GB

Test #2 :

File Type : 359 different files, Photos and Videos only (.jpg / .mp4)

File Size : Total of 860 MB

Test #3 :

File Type : 1000 text files (.txt)

File Size : Total 994 KB (~1 KB each)

Now to the benchmarking, I run each of these tests with each one of the different cluster size settings : 2, 4, 8, 16, 32 and 64 KB. I noted down the total time taken, then got the average speed by dividing (Total File Size in MB / Time in seconds), and then I checked the total disk space taken in the USB Flash Drive's properties. Total time is considered to be starting at the time the button paste is pressed, to the time the copy dialog dissappears (as in real life's every day copying). Here are my results :

Test #1 : Total disk space taken and file size was the same in all.

( We copied only 1 big file so it's not something unexpected)

FAT32 - 2K - 4:24 = 264 sec = ~ 4.54 MB/sec

FAT32 - 4K - 4:19 = 259 sec = ~ 4.63 MB/sec (Default)

FAT32 - 8K - 4:19 = 259 sec = ~ 4.63 MB/sec

FAT32 - 16K - 3:32 = 212 sec = ~ 5.66 MB/sec (Fastest)

FAT32 - 32K - 4.16 = 256 sec = ~ 4.69 MB/sec

FAT32 - 64K - 4:15 = 255 sec = ~ 4.70 MB/sec

As we can see.. when copying a big file in relation to the cluster size, we reached the highest speed using 16KB cluster size.. But overally higher cluster size got better results in this test. No big difference though. Only 16KB made a nice difference. I ran this test several times and got the same results.. Next one is the following :

Test #2 : 359 Files - Images & Videos - Total File Size 860 MB

Fat32 - 2K - 9:46 = 586 sec = ~1.47 MB/sec

*Disk Space Taken : 860 MB

Fat32 - 4K - 8:40 = 520 sec = ~1.65 MB/sec

*Disk Space Taken : 861 MB (Default)

Fat32 - 8K - 9:56 = 596 sec = ~1.44 MB/sec

*Disk Space Taken : 862 MB

Fat32 - 16K - 8:22 = 502 sec = ~1.71 MB/sec

*Disk Space Taken : 863 MB

Fat32 - 32K - 6:37 = 397 sec = ~2.17 MB/sec

*Disk Space Taken : 866 MB (Fastest)

Fat32 - 64K - 8.43 = 523 sec = ~1.64 MB/sec

*Disk Space Taken : 871 MB

Now this is the actual speed we get most of the times we copy files.. It is about files we copy-paste every day.. What's more usual than music, photos and videos.. up to 10 MB each.. Once again the default cluster size is not the best.. Best results here came from the 32K setting which I adopted for my personal USB flash drive.. Note that with bigger cluster size raises the Disk space taken a little bit.. Last test for today is the one with the text files.. The ones we think of less because they are so tiny... (~1KB..). Who cares about those right? Here are the results...

Test #3 : 1000 Files - TXT only - ~1KB each - Total 994KB

Fat32 - 2K - 1:17 = 77 sec = 0.0129 MB/sec

*Disk Space Taken : 2.07 MB (Fastest)

Fat32 - 4K - 2:02 = 122 sec = 0.0081 MB/sec

*Disk Space Taken : 4.02 MB (Default)

Fat32 - 8K - 4:34 = 274 sec = 0.0036 MB/sec

*Disk Space Taken : 7.93 MB

Fat32 - 16K - 9:47 = 587 sec = 0.0017 MB/sec

*Disk Space Taken : 15,7 MB

Fat32 - 32K -15:23 = 923 sec = 0.0010 MB/sec

*Disk Space Taken : 31.3 MB (Extremely Slow)

Fat32 - 64K - 7:44 = 464 sec = 0.0021 MB/sec

*Disk Space Taken : 62.6 MB

See what I'm saying..? Speeds are pathetic when it comes to such many files even at such a small size like text files.. It was really exhasting to complete all of these.. consider doing that 15:23 duration about 3 times to make sure everything is correct... Moreover, the disk space sacrificed for text files is also too big when we set the cluster size to higher levels. Think of that.. 1 MB of data takes 64 MB of disk space! Well, the first thing that comes to my mind (and probably yours) is that "I've never copied so many text files.." and after doing these tests.. I probably never will...

In conclusion, I guess it is a matter of which file type you are mostly using the USB Flash Drive for.. I mainly use it to move files 1~10 MB large so I guess I'll just stick to the 32K cluster size setting. Also I might format it to 16K in case I need to carry over a movie or something huge like that... The good thing is that the format process only takes less than 30 seconds.. So you might wanna switch the cluster size according to what you wanna use it for, each time.. The choice is yours.. I did the research for you and I hope I helped! Thanks for reading and don't forget to check back on from time to time..!