Haskell, TPS and performance

I am not a Haskell programmer but the concern is if the language really performs. So, besides the language performance, what is the current TPS provided by the network ?

What kind of Dapp will you be looking to deploy?

1 Like

I’ve never heard that performance is a big issue with Haskell, but I’m not a Haskell dev yet. A quick search suggests that it depends on how you code:

1 Like

Here is a good article on Cardano TPS:



thank you. My concern is… at this stage I was expecting Cardano team really publish real reports based in the network tests done so far… it is easy to estimate 5k TPS in a page but I would like to understand if all my investment is going to a real thing and not an unrealistic academic theory.

1 Like

It’s going into academic theory then the real thing in that order. :grinning:


so do I but the github activity seems to be very low regarding the protocol. At this stage we should know some benchmark.

Not sure which repo you are following, please check this site for an overview of updates across repos.


Truth is, nobody should rely primarily on the projects to measure performance. Are there any resources or references with respect to chain performance? It would be nice, but likely it is limited at this point.

From a systems engineering point of view, you need to develop models that are sufficiently detailed to simulate. For systems like this at scale, simulation is the only option besides evaluating the system in operation at full scale. You can get results from test systems/network, but there too you need good models to project from the test configurations to scale.

On thing we know is that Bitcoin does not scale and no one has proposed anything in that space to fix that. How can any global system be useful at 6-7 TPS?

Also, TPS will not be the only important measure. The important measure for clients is how long from when my transaction is submitted to a node until some level of confidence is reached. In principle, I see no reason the TPS should not be able to scale with traffic in the right architecture, but it may take longer to establish confidence. Also, some architectures may be able to establish a more limited confidence that assures a condition sufficient to verify conditions and claims WRT contracts. Not saying Cardano has this property, but it could related to flexible scaling in some architectures for DLT.

Gerry, usually during the development time in parallel test team produce artifacts to test the system. Sometimes create their own tool similar to JMeter and the performance must be defined as part of the system requirements that we call KPI (Key Performance Index).

I would like to understand if Charles and the talented team are doing it. There is no question about the talent of the team, but I have questions about the real SDLC (Software Developement Life Cycle) and how things are being managed.

1 Like

I think we pretty much agree. I’d like to see more models, test tools and moves toward more tools from complex systems engineering practice. I see their work on formalisms as a step that direction at the protocol level. I don’t think this part of the systems development practice has that many well developed tools and methods, so it is a challenge to say the least.

What I’m also suggesting is that if the issues are important enough to you, maybe you could contribute in this area. Even if the test tools are limited, you might be able to cobble something together based on what is there. Have you tried to find out what they use in the dev channels?

Yes!!! that’s my plan… this week I got my C4 certifications and I am completing my Haskell course… I know C, C++, Java, JS, Node, Assembly, Lisp, and other imperative languages but functional programming is something completely new to me… I would love to give my contribution not only on tech side but the development strategy… I am a technical program manager in a big company and I have projects with several moving parts in firmware, cloud (infra, front and backend) and mobile development.


The programming language used in has nothing to do with TPS. It’s based on the consensus algorithm that can be implemented in any language.