6500 x 600 x 150 bytes on average that would

Question

6500 x 600 x 150 bytes on average that would

make 585,000,000 bytes per 10 minutes, right?

Would that mean potential 600MB max block size amrite?

You also need to benchmark big blocks processing tho, probably.

#bch #bitcoincash #cryptocurrency #english

0

02.04.2024

10 ответов

21 просмотр

Shadow Of Harbringer 👁️🧀 Автор вопроса

tl121x
That might be ok if you were running a one person ...

Стикер

0

7 месяцев назад

minisatoshi

More than 20x lol

0

7 месяцев назад

matricz

tl121x
That might be ok if you were running a one person ...

If you want to build your infrastructure on RPis... you can.

0

7 месяцев назад

Shadow Of Harbringer 👁️🧀 Автор вопроса

matricz
If you want to build your infrastructure on RPis.....

That's the spirit!

0

7 месяцев назад

tl121x

The word I heard was that the pi5 performance was about 2x the pi4. So that the Pi5 should be good for 500MB blocks for general use. There are other machines only a little more expensive than the Pi 5 that are much faster and would handle GB blocks. Current node software single threads the UTXO database, with the major bottleneck being IO ops, perhaps 8 random IOs per transaction. Current NVMe SSDs are plenty fast enough but only with a queue depth of 32, otherwise IO latency is a factor in node throughput. My concern is the apparent lack of interest in node performance in the BCH community, probably because most developers don’t come a service provider or computer manufacturer or system integrator background. I think the bitcoin space lost this when Mike Hearn was driven out in 2016.

0

7 месяцев назад

Rock Dee B4sti0n

tl121x
The word I heard was that the pi5 performance was ...

What would you recommend that's slightly more expensive than Pi5 but are much faster? Thanka

0

7 месяцев назад

im_uname#100🍋 🌭

tl121x
The word I heard was that the pi5 performance was ...

i mean the reality is we've been fighting a survival war for years and it's kinda hard to "show interest in node performance" when we have 100kb blocks and a shoestring budget. naturally getting those economic activity in the first place would come before accommodating those economic activity.

0

7 месяцев назад

Remora_101🟢

tl121x
The word I heard was that the pi5 performance was ...

Didn't BCHN do a massive performance improvement right after the BCHA fork? Surely there is always more to do, but it's not like nothing has been done.

0

7 месяцев назад

Fernando

Remora_101🟢
Didn't BCHN do a massive performance improvement r...

I guess the only performance improvements were done in some mining-related RPC commands

0

7 месяцев назад

tl121x · Accepted Answer

That might be ok if you were running a one person private node. If you were running a business node or a public SPV server it would not be ok. At the very minimum you would need a factor of two safety factor to provide acceptable performance (queuing delays) during periods of peak load. There are also extra potential overheads associated with orphan processing. Also, resources are required for maintaining network connections to many neighbors and there can be considerable overhead in network traffic associated with mempool flooding of advertisements of new transactions. Worse, the node performance benefits from caching of chainstate, and will probably deteriorate during periods of temporary overload. This may depend on how your SSD is connected to the Pi5. In my experience architecting, building and measuring network based computer systems, I have found the best approach is to test systems to the breaking point and see what happens, and do so under all edge cases that are reasonably possible. Lacking that, there is always the possibility of over engineering by a huge safety factor.

209 похожих чатов

6500 x 600 x 150 bytes on average that would

10 ответов

Похожие вопросы