To minimise terminology confusion (transfer block size, filesystem block size / NTFS cluster size, RAID block size, etc.), let's use the proper term here, which is "stripe size".
You'll need to find out for yourself what gives you the best results. Take a few hours to do full benchmarks with the stripe sizes available to you. If I had to make a recommendation, it would be a stripe size of 128KB. The reason is that Intel SSDs tend to perform best with 64KByte transfer block sizes, so combine the two and you have 128KBytes. However, I won't be surprised if 256KB performs better. I wouldn't bother testing 512KByte; for 2 disks this is overkill.
If you're using OS X, I can't recommend any benchmarking applications. If you're using Windows, for benchmarking applications that are useful in this configuration, I would recommend HD Tune Pro (Benchmark tab, File benchmark tab, Random access tab, and Extra tests tab) and ATTO (transfer sizes ranging from 4KByte to 1024KByte are sufficient; you might also test transfer sizes of 512 bytes since that's what the advertised logical and physical "sector size" is on SSDs).
Before doing any benchmarking, make sure your partition is aligned to a 8KByte boundary (65536 bytes is a good choice). You can verify alignment under Windows using AS SSD (if it says "xxxK OK" then it's aligned, if it says "xxxK BAD" then it's not aligned).
Be sure to do a Secure Erase of both SSDs between each benchmark test. Failure to do so may skew your results depending on how much data was written to the SSDs during write tests; wear levelling can impact performance over time. You'll be spending a lot of time unplugging SATA power cables while running the Intel SSD Toolbox, so keep that in mind.
Be aware that if you're using Windows, a RAID configuration makes you lose TRIM capability; Intel's RST drivers, at this time, do not handle TRIM when multiple SSDs are configured in a RAID set. (TRIM is supported in RAID mode if each disk is JBOD / a single disk and not a RAID set). You'll be relying 100% on the GC (garbage collection) model in the SSD firmware itself. It performs well/decent, but TRIM is always better.
Also please be aware of the risks of RAID-0; the fact you're using SSDs doesn't minimise this risk in any way shape or form. If you have issues with either disk, you will lose your entire array. Do backups regularly. Mark my words, SSDs fail just like MHDDs do. Don't think you're safer because you're using SSDs.
Thank you koitsu. My apologies for the improper terminology but that is what OSx calls it in their drop down menu "block size". I know of a few mac benchmarking untilities for mac that I could try. I wanted to minimize writing to the drives since they have a finite amount of write cycles. I thought that there was a throw down strip size that I could use that would satisfy a RAID 0. As for the Trim support Apple does not support TRIM in any aftermarket drives on their laptop. If it isn't the SSD that came with the computer then TRIM will not work. This has been confirmed in numerous Forum, therefore since I can not take advantage of TRIM I will just do the RAID 0 and rely on the garbage collection. Commenting on your statement "Mark my words, SSDs fail just like MHDDs do. Don't think you're safer because you're using SSDs." I know i'm not safer, i've read the horror stories all over the internet of drives failing and consumers who have had their drives for 3+ years after all of the write cycles have been exhausted they can only read from the drive and their Laptop/Desktop/Notebook/Netbook becomes a lemon. I do understand that fully. I did some research on Intel 320 garbage collection and it seems pretty good compared to the X25's and other SSD's if been reading up on. The thing is I will not know because only time will tell. I do video editing (rendering and reencoding) a lot and I need that improved performance that the SSD's can offer, this is why I chose the SSD plus I needed the space which, there was only one choice 600gb (since it's the biggest size). Referring to your comment "(TRIM is supported in RAID mode if each disk is JBOD / a single disk and not a RAID set)" There is an option when setting up the hard drives called Concatenated (which I believe is the same as JBOD). The reason I do not opt for that is because you do not get any speed improvement using this option and speed is the key reason why i'm doing RAID 0. I think I will just take your educated guess and set the 'stripe' size to 128kByte. If I have the time to test it out I will just test 128kByte and 256kByte with the benchmarking utility and see which one gives me the best performance. I do have one last question, I just looked it up and this is in reference to your statement "To minimise terminology confusion (transfer block size, filesystem block size / NTFS cluster size, RAID block size, etc.), let's use the proper term here, which is "stripe size".", isn't RAID block size and stripe size the same thing?
I thank you for your prompt response, you have helped me a lot, I do have to do some research on aligning my partition (not sure how to do that). You said "Before doing any benchmarking, make sure your partition is aligned to a 8KByte boundary (65536 bytes is a good choice). You can verify alignment under Windows using AS SSD (if it says "xxxK OK" then it's aligned, if it says "xxxK BAD" then it's not aligned)." I found an article on wikipedia explaining it - http://goo.gl/qHPJl - but I thought that was actual how the structure of the memory was. I thought NOR vs NAND aligned their memory in a certain way which gave you pro's and con's depending on what you would use the memory for. I guess I am not understanding the alignment to well. I'm not to worried as you said AS SSD would be able to tell me if it's ok or not, well i'm sure that the MAC equivalent to that benchmarking utility will tell me as well.
Once again I thank you for your response and your detailed explanation. I really appreciated it. Enjoy your day.