Howdy,

I want to dive in and address some quick points on how we think about performance and what you can expect from launch. First let me just say that we see launch or 1.0 as the starting line. We're in this for years to come. Perf is part of that commitment.

Performance is a wide range of topics but in terms of top priorities for perf we have 3.

• Accessibility : We want people with low end computers to be able to join their friends in VALORANT. So invite your buddy with his Intel HD 4400 to game with you. They won't be slamming 144FPS down but hopefully they can have fun. For the most part, we seem to be hitting the mark here, there is ongoing work to get all character ability VFX in budget here but it's coming along at a fast rate. We improve a number of abilities every patch. Patch .50 fixed a rather large issue introduced with .49, one of our key GPU optimizations regressed and we had to go fix it up again.

• ...

Howdy,

That graph definitely isn't normal. Each graph point is 1/10th of a second. You can see in some cases you're losing ~19% of your packets in those periods. This would likely lead to enemies sometimes being in the wrong position or in more severe cases you might rubberband back to a previous position. Your shots are likely to be delayed as well.

Really you want to see the packet loss graph have 0 spikes in it if your internet is working properly (or maybe some smaller spikes every now and then, but really you should expect to see 0%).

If you're on Wifi you might try a wired connection. If not, you might need to restart your modem or router. If the problem persists you might try reaching out to player support or even your own ISP to troubleshoot the issue further.

I had a problem like this earlier today and was able to fix it by fixing some loose cables on my router.

The menus are very very easy on the CPU. So you end up GPU limited in the menus (GPU usage goes to 100% if you are uncapped in menus). When you load into game there is more CPU processing that has to be done to deal with processing the actual game. At that point you become CPU limited on a lot of newer systems. Due to the fact that the CPU isn't generating enough frames your GPU load goes down below 100%.

I'm a little surprised bind is the worst performance for you. I think the order we have is:

1 Haven

2 Bind

3 Split (Hasn't had a full optimization pass yet)

The maps will never be 1-to-1 in terms of performance due to being different maps and different hardware always produces different results but it's something we're keeping a close eye on so thank you for the feedback.

Ahh yeah that's cool. I don't think you're trying to push a false agenda or anything. Just trying to stay accurate.

I can already see some people parroting "single-threaded" around so I wanted to make sure people understand that a quad+ core CPU is still a really good idea for VALORANT since you can think of it as a 3x threaded game which is a big difference from a 1x threaded game.

If we want to be super pedantic I VALORANT allocates ~1 worker thread per virtual core on your system. However, those threads don't always have work to do and can spend a lot of their time idle.

For sure! I considered adding dual channel memory to our min and recommended specs. We see the same thing internally where the game won't run well on one stick of RAM. It seems like it may be more popular to run just one stick of RAM than we originally thought. Dual channel RAM is going to speed a computer up a ton and not just for VALORANT.

The ping chart averages over 1/10th of second. I think the scoreboard is 1/4th of a second. It's possible you're seeing really transient ping spikes that are actually much higher than the chart shows.

Could also be packet loss so you might try turning that chart on as well.

Cmon Riot, it's so f*cking SIMPLE!

.RunValorant(parallelize = true, threadCount = THREADS.MAX, coreCount = CORES.MAX) parameters, sh*t, how did the devs forget those options?

Ahh sh*t you're right I forgot.

On a more serious note though, we do hold ourselves to a high perf bar for the game. I'm glad players are giving us perf information from the more subjective side. We do have numbers but hearing from player's experiences directly as well is great, both when they're having a great performance and when they think we're missing the bar.

I just want to jump in and clarify that our game is indeed multi-threaded. The game is using 3+ threads nearly 100% of the time. It varies based on the amount of work that can be multi-threaded at the time.

My post was to clarify that on higher end systems the best way to improve framerate is to increase core-clock of the CPU and to let people know upgrading their GPU wasn't going to help as much if you already have a pretty modern GPU.

That being said we can always look at multi-threading more of the systems to get even better performance on systems that have a large number of cores. It's something we're aware of.

Tuesday is the plan for the next patch barring anything that would delay it.

Split is still in a beta state (shipped early to increase map pool for players). We have updates in the next patch that should improve perf for that map.

We have seen that Bind is lower performance than Haven. The maps will never be exactly equal in perf but we're keeping an eye on Bind to see if adjustments need to be made.

The patch did have an overall perf decrease of about 10% on average (varies from person to person). We've already identified the root cause and the next patch will return that perf to players.

It's a bit unrelated to this post but yeah. I don't think average FPS will get much better on that hardware but we are aware that certain character abilities or effects drop the FPS below our acceptable bar on that hardware. We're working through those specific optimizations each patch.

If I'm honest this is one of the places where covid-19 has delayed us the most. We don't have hardware labs in our homes unfortunately (and even if we did, we don't have the space to support them). Testing hasn't stopped here. It's just significantly longer turnaround time.

Just want to hop in to say hi we see you!

BIG CAVEAT: We don't have all our analytics around client perf data set up perfectly yet.

That being said I can see the frametime dipped about 1ms (or about 10%) on average after the update. There could be many explanations for this, the simplest being we've made a mistake in one of the changes we made in the update. We're going to investigate all angles and figure this out as quickly as possible.

Based on the reports we're seeing and our own experiences internally we believe it could affect different users to different degrees so everyone's experience might not be the same. Try to be empathetic to other players whose experiences may legitimately not be the same as your own!

One interesting thing we're seeing in the data is that it seemingly affects all parts of the game simultaneously (CPU-side). This narrows down our suspects considerably but none are clearly smoking guns a...

Player support is a better place to get personalized support. They're more familiar with a wide variety of systems and problem-solving. You can create at ticket here: https://support-valorant.riotgames.com/hc/en-us/requests/new

Since you mentioned others with the same CPU get better performance off the top of my head I'd suggest:

Checking CPU temps (can be down-throttled if it gets too hot) Checking that your RAM is installed in the proper slots to make use of dual-channel. (this can be killer to performance and is hard to detect)

However, Reddit isn't really a great place to ...

The difference between a gamer and a casual is that when gamers facepalm into their keyboards their face is colorful, it's also more clicky.

Yeah I share you opinion here. It's definitely a balance of making it understandable + giving players control.

I do really read this! I don't work on the battlepass stuff but I'll pass the feedback along.

"Is valorant more CPU intensive or gpu intensive?"

I would think about it like this:

If you go out and buy a brand new top-end GPU/CPU today. Your CPU is going to be the limiting factor. We'll probably only use 30-50% of the GPU.

If you have a mid-range computer that you bought 4 years ago. You're probably close to balanced.

If you have a mid-range computer 4+years ago you're probably going to be GPU limited.

We expect that about 50% of players will be CPU bound (higher-end machines) and 50% of players will be GPU bound (lower-end machines). However, this changes as we make specific optimizations for CPU or GPU.

We're looking at what we can do there as well. It would be a separate change from the two changes I mentioned.

What would you like to see there personally? Should be tied to muting someone on Voice? Should it be it's own separate mute?

Disclaimer: I'm a Software Engineer not the one who's making UX decisions but I have chatted with them about it.

WE LOVE THE GOOD BUG REPORTS! Keep em coming Reddit.

Yeah something like this could definitely work.

However, it might have an unintended effect.

Imagine you stored 99% of a "down movement" locally on the client then try to hit a flick really fast. The mouse movement for the flick would need to overcome that 99% "storage" value to work. Might lead to some inconsistencies that are hard to detect but throw of the "feel" of the game if that makes sense. You could counter this in various ways "hand waving" so the basic idea could work with the right implementation.

When we look at problems like this we often write down all possible solutions and share them amongst the dev team to try and find the most optimal solution for the player experience. I'm sure something like this will be one of the options on the docket. The easiest solution is probably what RiotStealthyJesus suggests, simply increasing the precision to the point where it works for all sens and zoom factors.