Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Ryzen browser benchmark - how does it compare to Atom?

This site may earn commission on affiliate links.

Mrklaw

Active Member
Mar 5, 2020
2,623
1,734
Berkshire
My M3LR has the ryzen processor, so I went to www.browserbench.org and ran the Jetstream2 and Motionmark benchmarks on the built in browser.

Can someone with the atom CPU do the same and we can compare notes? Maybe this isn't that useful - eg it may not be able to access the GPU in the browser, so it will be CPU bound, and it might need apps to do that, but I was curious :)

These were on a non-maximised browser window
Jetstream2 : overall score 63.534
Motionmark: overall score 264.73

screenshots showing the different benchmark task scores in case it shows anything interesting

5759689B-D3D1-4C44-8B1E-67E8DDB27BA5.jpeg


C8FE70CE-BCAF-4C50-A671-F61254CEB92D.jpeg
 
I’d love to see the results of an Atom to quantify what I am experiencing. Having moved from an Atom Model 3 to a Ryzen Y last week, I am totally blown away by the speed increase. Things like the browser, YouTube etc are so much more performant. It’s night and day.
 
What about normal UI use whilst driving? I’m guessing there’s no difference as this isn’t too intensive?
I never found that to have any lag before. My perception of the normal UI was that it was always snappy. Perhaps things like pinching and zooming around a map are marginally quicker but not like the step change in performance of the browser based apps.
 
  • Like
Reactions: Lord Farquad
Perhaps things like pinching and zooming around a map are marginally quicker but not like the step change in performance of the browser based apps.

We hade a new "car computer" AKA MCU2 last September. It should be like for like (especially as most things were backed up from old and carried over giving an otherwise seamless fix), but who knows if any minor revisions took place in replacement units in the 2 years since we got the car. I think there were some production changes to IO / ecall etc, but they were quite fundamental so without other changes I suspect they wouldn't have made it to replacement units. That may explain why we got our replacement so quick and others had to wait for months.

After the new unit, I noticed that map loading wasn't so snappy as before, which were always pretty instantaneous. I had assumed that it was going through the process of caching the map tiles, but never quite sure, especially as we had a software update (2021.32.x from 2021.24.5) whilst car was in service. I've not paid any attention to recent speeds to know if its still a bit laggy on pinch and zoom. Easy to over think the normal things after something has changed. It possibly did it all along, but I just never noticed.
 
Didn't have much luck with Jetstream2 - got part way through, probably 5+ mins and then white screen and back to start test again :( - it was going very slowly. I imagine the new system has more memory as well as a faster CPU.

Motion Mark 13.31 - LOL, so new processor with a score of 265 is much better although to put it into context my iPhone got 906 and my laptop 1752 without breaking a sweat or consuming more than 10W on average (27W peak) including the power for the screen, so it's hardly a high performance CPU whatever is in there.


1645815827355.png
 
interesting about jetstream. It did take a while on my car while parked up. iphone is pretty powerful but perhaps the browser can’t tap into the GPU? Still seems a good lift above the atom - either CPU, ram or a bit of both.
 
I'm not convinced it's worth the range loss, unless you use the apps a LOT. The low voltage battery in the new car is awesome though

I’d be curious about that too - wonder if anyone is able to monitor power usage of the MCU. 12 miles is a lot, which suggests much more wattage for the MCU. That might tally with the CPU running at high speeds and the discrete GPU, but maybe thats only when taxed? a lot of the time it may be runnign slower, or not using the GPU at all. The basic software shouldn’t be any more demanding and Ryzen should be an efficient chip, so shoudl be ‘racing to idle’ not burning watts.

the higher performance shoudl be there for more headroom, without being less efficient for normal tasks (at least not *that* much less efficient)
 
12 miles is a lot
It's a ridiculous amount - over 3% of the battery or around 2.5 kWh.

Given a 360 mile range and a WLTP average speed of 29 mph, the vehicle would need to be on for 12.5 hours, therefore a constant 200 W over 12.5 hours to consume the 2.5 kWh.

200 W more than the old system! That cannot be right, especially when you are driving you aren't making any use of the extra power so the processor should just be idling.

I wonder if running the low voltage electrics at 15.8 v rather than 14.4 v is also contributing to more power being consumed.


Maybe Tesla are crypto mining in the background to secretly increase their revenue 🤣
 
  • Funny
Reactions: RichardTheKing
It's a ridiculous amount - over 3% of the battery or around 2.5 kWh.

Given a 360 mile range and a WLTP average speed of 29 mph, the vehicle would need to be on for 12.5 hours, therefore a constant 200 W over 12.5 hours to consume the 2.5 kWh.

200 W more than the old system! That cannot be right, especially when you are driving you aren't making any use of the extra power so the processor should just be idling.

I wonder if running the low voltage electrics at 15.8 v rather than 14.4 v is also contributing to more power being consumed.


Maybe Tesla are crypto mining in the background to secretly increase their revenue 🤣
I also wondered if this 3% loss was a just in case scenario and after monitoring the situation for a few months that the range will be restored
 
I also wondered if this 3% loss was a just in case scenario and after monitoring the situation for a few months that the range will be restored
Yes, as it seems too much just for a MCU update.

Maybe they didn’t have time to do the full WLTP testing so they lowballed it so they are definitely not overstating range based on the maximum power that the MCU could consume, not what it actually consumes. Or there are additional changes that have contributed which we don’t know about. Or the new 79 kWh battery isn’t giving quite as much range as they initially thought.
 
Yes, as it seems too much just for a MCU update.

Maybe they didn’t have time to do the full WLTP testing so they lowballed it so they are definitely not overstating range based on the maximum power that the MCU could consume, not what it actually consumes. Or there are additional changes that have contributed which we don’t know about. Or the new 79 kWh battery isn’t giving quite as much range as they initially thought.
Interesting point....does anybody know whether the new chip is going in the M3 RWD and if yes are they reducing the range of the LFP battery ?
 
Interesting point....does anybody know whether the new chip is going in the M3 RWD and if yes are they reducing the range of the LFP battery ?
Yes, the first ones have appeared in Australia. Range is the same as Europe and wasn’t lower but perhaps they knew the new processor was coming and factored that in when they switched to the new 60 kWh battery, or it is indeed just related to the new 79 kWh battery not being as big as they first thought.
 
  • Like
Reactions: Bouba
Didn't have much luck with Jetstream2 - got part way through, probably 5+ mins and then white screen and back to start test again :( - it was going very slowly. I imagine the new system has more memory as well as a faster CPU.

Motion Mark 13.31 - LOL, so new processor with a score of 265 is much better although to put it into context my iPhone got 906 and my laptop 1752 without breaking a sweat or consuming more than 10W on average (27W peak) including the power for the screen, so it's hardly a high performance CPU whatever is in there.


View attachment 773710
Although it might sound like a logical argument, it turns out that comparing your Tesla to another car or device isn’t going to provide a fair comparison. The problem here is that the performance of benchmarks like this comes down to support for hardware acceleration - clearly the Tesla browser doesn’t have it, therefore comparing to a device where hardware acceleration is available in the browser isn’t going to produce a like for like comparison. To emphasise this point, my phone (iPhone XR) gets 965 on Motionmark and my phone definitely isn’t 72 times faster than the Atom processor. I think the only thing this benchmark is useful for is to compare the performance delta of the Atom and Ryzen processors when running the Tesla browser.
 
  • Like
Reactions: Mrklaw
I tried Silverbench (seems to be a javascript based benchmark kinda similar to cinebench) on my car (Ryzen M3P) and my dad's car (Atom MYP). By that, the Ryzen is ~5x faster, which seems about right when looking at benchmarks of the two CPUs in other devices

The actual numbers are low compared to what you'd get on other devices - but I think it's as others have noted: the Tesla browser is poorly optimized and doesn't have much in the way of hardware acceleration

Don't have a great way of truly evaluating testing the GPU performance. Doesn't seem like the browser can kick the Navi23 on.
 

Attachments

  • IMG_2937.JPG
    IMG_2937.JPG
    818.2 KB · Views: 34
  • IMG_2981.JPG
    IMG_2981.JPG
    592.7 KB · Views: 29