anyone know how the TPS stats look like at the moment? In theory it should be unlimited for SAFE right? In last week update they wrote> ,so that the artificial limit on the number of transactions per second that we have at the moment may be lifted, .
The last Parsec optimization is here, but keep in mind that the “unlimeted” number of TPS in SAFE comes from the fact that the network is fully distributed, not only decentralized as other solutions, so the number of TPS increases as the number of vaults, and sections, grows.
Of course as Parsec is improved and optimized the TPS per section, and globally, will be higher.
4 Likes
can you see any number of TPS provided there? not sure where to look for it, too many different stats there…
It’s a bad measure when batch processing is involved. So you can batch 100 million transactions in a block and get consensus after 5 seconds (super dead slow consensus) and it appears like thousands and thousands of TPS. You can measure them all with a single transaction per block, but honeybadger etc. would look like treacle.
tl;dr Currently reporting TPS accurately is a bit of a magic trick in all the current BFT and ABFT developments out there. We prob could issue tests and figures for 500,000 tps for instance but it’s not really honest if the network waited for 500,000 transactions to process them like that? Then the speed is not PARSEC alone but network queues and what size batch to process. Basically with batch stuff if the batch is small the TPS is low, if it’s high then … Hope this helps (don’t be fooled by published super amazing TPS figures).
9 Likes
Ok thanks, I thought the TPS will be revealed at some point, to ilustrate the advance tech and high performance that SAFE will provide to the people when comparing to blockchain projects…
@digipl does that mean that the perf task the team was hinting at is now complete and merged?
PS. I think we can expect the Parsec Whitepaper v.2 to be released on Friday at latest.
1 Like
I see ,implement mock Parsec, in the routing github repo, are we getting testnets any day now?
2 Likes
I believe soak tests run internally, have to come first.
2 Likes
Ah ok, I thought we could get a big update today…
1 Like
Yeah, thought so too, thought we get the first testnet today.
Even though we won’t get test-nets today, it feels like it is getting closer and that it isn’t far away, that is a good feeling. The champagne is on cooling and ready for when it happens. 
8 Likes
I understand your reticence to participate in the number padding game when it comes to tps reporting. Still, I strongly think that PARSEC will benefit from a marketing perspective from having a tps number attached to it. Just be very clear about how the tps number was determined, under which conditions, etc.
It’s just much easier for someone writing about consensus algos to add PARSEC to such list if it has an advertised tps benchmark. If it doesn’t, they likely would not even include it to begin with. That’s human nature. They won’t do the hard work. A tps number, no matter how imperfect, would be a heuristic for quick speed comparison with others for such a person.
11 Likes
Does anyone know how much is still to be done with the Malice part? From all the Alpha 3 pieces which need to be completed: dynamic membership, malice, testing (test nets), documentation, I believe we can tick off the first one as its done right? Next up is Malice, which i though is on brink of beeing finished, but seems like I was wrong or?
I think when Malice is done, we will see some soak tests being run. I think a couple steps from the second task have bleed into the third and now they are being finished in tandem with Malice, in a race to Alpha 3.
2 Likes
So this means there is no chance to see a test net this week? What would you guys say?
Never say never, but very doubtful.
2 Likes
thanks for the details, lets hope youre right, surprised the team doesnt respond, like its some kind of secret…
Or they are busy, you are asking during normal work hours 
12 Likes
Arghhhh, you were going so well with your questions up to then.
Ah jks, it’s kinda like the little tacker that learns if they keep yelling louder and louder eventually mum or dad give in. In this case twisting the language 
in ever so softly under the guise of seemingly relevant questions in an effort to squeeze out what you want. You do it really well tbh.
It’s an effective technique against honest people, unnecessary but very effective I’ll give you that
Ps: I’m old and don’t do gentle so 
11 Likes
According to Jira there are just a couple of small tasks left to be done on the Parec Fleming repo, also in the routing there are some implementations going on- is it reasonable to think that the soak testing and testnets are just days away?
3 Likes
It is reasonable but you never know what pops up at the last minute.
3 Likes