Original author: @Justin_Bons
Original translation: Peisen, BlockBeats
The two biggest poster children for “fake” TPS metrics are Solana and ADA: Solana misled investors by 6.5x, while ADA misled by 26.5x. Both are guilty of ignoring industry standards for TPS, and it’s time to break down the numbers and separate truth from fiction:
First, here are the current facts:
Solana’s maximum theoretical TPS is 10,000; ADA’s maximum theoretical TPS is 18
Solana’s current real TPS is 739 ; ADA’s current real TPS is 0.4
Depending on who you ask, you’ll get wildly different answers: some claim Solana’s maximum theoretical TPS is 65, 000 , while others claim ADA’s maximum theoretical TPS is 477 .
There is a huge discrepancy here, and in order to find the truth, something has to be given, because none of them can be true. As the explorers give us, the numbers are very different:
First, we have to separate usage metrics from maximum theoretical capacity metrics. Starting with Solana’s maximum theoretical capacity: Solana’s compute limit is 48M CUs per block, and a basic 1-to-1 TX requires 450 CUs; therefore, at a 0.4 second block time: 48M÷450= 106k÷0.4 = 266K TPS.
However, in Solana’s case, we cannot take this limit at face value. Due to other cryptographic limitations, the actual bottleneck is significantly lower, and the lowest bottleneck now appears to be EDDSA validation, which brings the maximum theoretical TPS down to about 50,000.
@bw_solana identified these bottlenecks; ~250k TPS data ingestion, ~125k TPS signature verification @hdd_edy pointed out the ~30k limit for EDDSA, and then @mrJackLevin raised it to ~50k. Given the ~50k bottleneck, we deducted voting TXs (75%) and "failed" TXs (10%).
It would be unfair to deduct “failed” TXs without impacting users, as they are destined to fail and pay fees. However, for the sake of argument, it is still deducted, giving Solana a maximum theoretical TPS of 10,000.
Now let’s look at the maximum theoretical capacity of ADA: the block size limit is 90,112 bytes, and the basic 1-to-1 TX is 250 bytes; therefore, at a 20 second block time: 90112÷20 = 4505÷250 = 18 TPS. We will take this at face value, as there is no lower bottleneck than this.
One of the misleading parties is http://eutxo.org, as it provides shockingly "false" TPS numbers, while almost all other explorers outside of ADA report correct TPS numbers, such as http://chainspect.app, which has consistent and comparable TPS metrics.
So why does ADA claim to have 26x more capacity than it possibly has? (Compared to Solana’s 6 x) This is because ADA counts multiple outputs as separate TXs.
Almost all other chains are able to send in batches without increasing the cost. However, no one actually counts it in TPS, even BTC can do this with Schnorr signatures, yet no one seems to claim that BTC can do 400+ TPS?
Batching is a great feature to expect on all modern chains, and I'm not even opposed to including it as long as we do it consistently (x 20 TPS for all chains to accommodate ADA) However, at this point, It would make more sense to invent a new term, such as "output per second." Since batch processing does not scale the number of users, only the number of TXs is scaled.
Since all outputs still come from the same private key (just scaled with the custodian), this is where the TPS and “OPS” metrics may be valuable. Just like how “Exchanges Per Second” is a new metric to measure capacity in a more granular way. However, this is why saying that ADA has such a high theoretical maximum TPS is misleading, after all it is far from the truth.
This is all particularly ironic after ADA’s community constantly attacked Solana’s metrics. Because in ADA, the disparity between the ratio of what is claimed and reality is greater than in Solana. They use the exact same TPS methodology, even after deducting consensus and failed TXs.
Now let’s look at current usage, starting with Solana: Raw TPS numbers can be misleading because they include consensus information, which is why we should deduct for that, 3654-2740 = 914 x 0.90 = 822 TPS, I also deducted failed TXs purely for the sake of argument
This is why the wider adoption of the “True TPS” metric is such a positive development for Solana, representing a significant departure from previous eras. In a more misleading way, most Solana explorers now display “True TPS”.
When it comes to current ADA usage, we can again take the 0.4 TPS at face value as there is no reason to calculate it differently. The fact is that almost no one is using ADA and that is what TPS represents as most other cryptocurrencies have far more usage than ADA.
Capacity and scalability are two different things, at least in computer science. We might be able to view ADA more favorably from this perspective: Since scalability is about comparing node requirements to capacity, however, since Solana has 555x more capacity, this isn’t a fair comparison. Even the difference in hosting costs for ADA’s low node requirements and Solana’s high node requirements doesn’t come close to this capacity gap.
Technically, this makes Solana more "scalable" than ADA, even if that's an unfair comparison since ADA doesn't even seem to be trying. After speaking with the community, they seem to think that when usage comes, they can increase capacity. However, my stance is that if they keep capacity low, usage will never come, which is why ADA has failed to attract more usage!
ADA has missed wave after wave of adoption due to this. Please don’t blame me as a critic trying to save a large community from becoming increasingly irrelevant as the ADA community has become highly closed and hostile to outside critics.
There will be a lot of accusations about my motives because I am a neutral party, but I have never accepted any form of compensation, never received a token allocation, never even accepted an advisory position, and gave up millions of dollars to keep my credibility intact. The investment firm I am affiliated with, Cyber Capital, can convert its entire SOL position to an ADA position if it sees fit.
ADA clearly has a large community that does care about decentralization. Sincere disagreements should arise from the modular vs. monolithic debate, as ADA ignores L1 scaling. ADA should focus more on L1 scaling, and the current maximum capacity of 18 TPS prevents builders outside of ADA from joining. ADA learned a hard lesson when the market chose Solana with its 10k TPS capacity despite Solana outages, node centralization, and past lies.
Original link
