BitShares will launch on schedule with enough throughput to support all current needs plus all of Bitcoin and most altcoins combined (if they wanted to move onto our platform for some fiendishly wise reason).
We are still optimizing the network code, which is not part of the blockchain itself, so that in coming months we can continue to scale up to full design specs of 100,000 TPS without even a hard fork.
The test net is where we are doing this optimization, so when you hear about a test "fail" is has nothing to do with what is launching on October 13. We will have many more such fails on the test net side as we tune up to light speed, but the production blockchain where everybody's funds reside will hum along without breaking a sweat.
We are still optimizing the network code, which is not part of the blockchain itself, so that in coming months we can continue to scale up to full design specs of 100,000 TPS without even a hard fork.
The test net is where we are doing this optimization, so when you hear about a test "fail" is has nothing to do with what is launching on October 13. We will have many more such fails on the test net side as we tune up to light speed, but the production blockchain where everybody's funds reside will hum along without breaking a sweat.
From what I read on the blog the benches were done without signature verification. Why would you bench without verification of signatures ?
The actuall speeds will prob be much less impressive unless I missed something (just skimmed the post) or all delegates have an absurd hardware requirements. Even then you'll have a great time putten all those data onto the wire. I'll believe those 100k TPS when I see them. Luckily we probably won't see them anytime soon since there isn't really a need for it.
Signature verification is parallelizable. The benchmark is for the One Thread that must remain serial. With that thread running at warp speed as demonstrated, we can add parallel feeders as the network scales and generates the revenues needed to pay for it.
I'll remind you of the quote often attributed to Bill Gates that, "640 kB ought to be enough for anybody".
You still could have benched it with signature verification (done in parallel) to give people an idea of the hardware requirements for such speeds. Now we know that you can process it on a mediocre CPU with signature verification disabled which really doesn't say anything about the real world and is imho mildly interesting at best.
