• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

SunSpider 1.0 JavaScript benchmark results

Joined
Jul 15, 2022
Messages
718 (1.07/day)
Capture d’écran du 2023-06-11 11-28-07.png

I score 113.5 ms

Hardware: Intel i3-3240 + 8GB RAM @1600MHz DDR3 dual-channel + NVIDIA GTX 650 1GB + EVO 850 500GB
Software: Firefox 113 + NixOS + Cinnamon




2023-06-11-142627_1920x1080_scrot.png


In this test of the world's most popular programming language, FreeBSD is significantly faster than NixOS.
Although NixOS is faster than 98% of all other Linux systems in this test.

You can take the test on this page. http://proofcafe.org/jsx-bench/js/sunspider-1.0/driver.html
When you post, please include your specific hardware and software.
 
Last edited:
Joined
Jul 15, 2022
Messages
718 (1.07/day)
Capture d’écran du 2023-06-13 12-06-52.png

Hardware: AMD R5 PRO 3400G, dual-channel DDR4 @2666MHz
Software: Epiphany browser 44.3 (WebKit version 605.1) + NixOS + Cinnamon
 
Joined
Jul 15, 2022
Messages
718 (1.07/day)
2023-07-19-182905_1920x1080_scrot.png

102.1 ms

Hardware: Intel i3-3240 + 8GB RAM @1600MHz DDR3 dual-channel + NVIDIA GTX 650 1GB + EVO 850 500GB
Software: FreeBSD 13.1 & Firefox 115.0.2
 
Joined
Jun 21, 2021
Messages
2,901 (2.74/day)
System Name daily driver Mac mini M2 Pro
Processor Apple proprietary M2 Pro (6 p-cores, 4 e-cores)
Motherboard Apple proprietary
Cooling Apple proprietary
Memory Apple proprietary 16GB LPDDR5 unified memory
Video Card(s) Apple proprietary M2 Pro (16-core GPU)
Storage Apple proprietary onboard 512GB SSD + various external HDDs
Display(s) LG 27UL850W (4K@60Hz IPS)
Case Apple proprietary
Audio Device(s) Apple proprietary
Power Supply Apple proprietary
Mouse Apple Magic Trackpad 2
Keyboard Keychron K1 tenkeyless (Gateron Reds)
Software macOS Ventura 13.6 (with latest patches)
Benchmark Scores (My Windows daily driver is a Beelink Mini S12 Pro. I'm not interested in benchmarking.)
Here's my daily driver with my usual browser poison:

js-bench-m2pro.jpg


Chromium dev build on Mac mini M2 Pro: 63.7ms

Hardware: Apple Mac mini M2 Pro, 6 p-cores, 4 e-cores (macOS Ventura 13.4.1), 16GB unified memory, 512GB SSD
Software: Chromium 115.0.5790.0 (Developer Build) (arm64)

This benchmark runs faster on Firefox on the same hardware:

js-bench-m2pro-firefox.jpg


Firefox on Mac mini M2 Pro: 48.4ms

Software: Firefox 115.0.3esr (64-bit)

I tried on the Mac's Safari.

js-benc-m2pro-safari.jpg


Safari on Mac mini M2 Pro: 39.3ms

Unsurprisingly Apple's own browser running on their proprietary hardware is king.

As usual, my iPad blows away my various Windows computers for this type of benchmark.

Sorry, no screenshot for the next one:

Safari on iPad mini 6th gen: 43.9ms

Hardware: Apple iPad mini 6th generation, 2 p-cores, 4-e-cores
Software: Safari (iPadOS 16.5.1)

iPad mini destroys everything on a performance-per-watt metric.
 
Last edited:
Joined
Jul 15, 2022
Messages
718 (1.07/day)
Unsurprisingly Apple's own browser running on their own proprietary hardware is king.

As usual, my iPad blows away my Windows computers for this type of benchmark.
The SunSpider benchmark was developed by Apple's WebKit team.
On the AMD 3400G PRO I see that Epiphany and Nyxt browser are faster than Firefox and Chromium-based browsers in SunSpider.
Epiphany and Nyxt browser use WebKit (like Safari and Orion browser).
 
Joined
Jun 21, 2021
Messages
2,901 (2.74/day)
System Name daily driver Mac mini M2 Pro
Processor Apple proprietary M2 Pro (6 p-cores, 4 e-cores)
Motherboard Apple proprietary
Cooling Apple proprietary
Memory Apple proprietary 16GB LPDDR5 unified memory
Video Card(s) Apple proprietary M2 Pro (16-core GPU)
Storage Apple proprietary onboard 512GB SSD + various external HDDs
Display(s) LG 27UL850W (4K@60Hz IPS)
Case Apple proprietary
Audio Device(s) Apple proprietary
Power Supply Apple proprietary
Mouse Apple Magic Trackpad 2
Keyboard Keychron K1 tenkeyless (Gateron Reds)
Software macOS Ventura 13.6 (with latest patches)
Benchmark Scores (My Windows daily driver is a Beelink Mini S12 Pro. I'm not interested in benchmarking.)
The SunSpider benchmark was developed by Apple's WebKit team.
On the AMD 3400G PRO I see that Epiphany and Nyxt browser are faster than Firefox and Chromium-based browsers in SunSpider.
Epiphany and Nyxt browser use WebKit (like Safari and Orion browser).

This is an excellent demonstration why I don't care much for browser benchmarks: each one favors a different browser engine.

When I browse the web, I end up on sites that use a wide range of different features, types of scripting, even code quality under differing circumstances. There are some websites that are reduced to helpless crawl when accessed on a slow and/or congested network while other sites run quite fine under identical conditions. Once congestion is eliminated, those pokey sites run fine.

It's similar to a single game graphics benchmark. Once again, each game title favors a certain combination of hardware (Intel, AMD, Nvidia, whatever) and software (DirectX, Vulcan, raytracing) technologies. A game that benches fine on an 8GB card at 1080p might suck at 1440p because 8GB of VRAM is insufficient. Other games might favor larger memory caches, more memory bandwidth, etc. With super sampling technologies in play, image quality is not quantified in benchmark results. Newer technologies like DLSS 3 Frame Generation might smooth out image playback but introduces input latency. Et cetera ad nauseam.

A more useful analysis would comprise a suite of various benchmarks. Some PC/game sites actually do this, running 10, 15, 20+ game benchmarks for a more useful measurement.

After all, it's not like I play one game or visit one website.

I run benchmarks on my Windows PCs mostly to configure fan curves. But once those fan curves are set (usually on the hottest day of the year) I can generally forget about them unless I reconfigure the build.

And web browser technology evolves pretty quickly, both on the client and server sides. Games might be patched a few times after release but it's rare to see major performance changes after a couple months of patches.

From a real world usage perspective, these benchmarks are useless. Let's say Browser A runs best on Site X, Browser B runs best on Site Y, and Browser C runs best on Site Z. Joe Consumer isn't going to constantly switch browsers because one browser runs marginally faster than another on a given site. Joe Consumer will favor one browser and use that until they encounter a site where that browser is broken. Even then they may decide to forge onward with a mediocre site visitation experience, get what they need while suffering, then go elsewhere. There's no requirement to run the best browser for each and every website.
 
Last edited:
Joined
Jul 15, 2022
Messages
718 (1.07/day)
This is an excellent demonstration why I don't care much for browser benchmarks: each one favors a different browser engine.

It's similar to a single game graphics benchmark. Once again, each game title favors a certain combination of hardware (Intel, AMD, Nvidia, whatever) and software (DirectX, Vulcan, raytracing) technologies. A game that benches fine on an 8GB card at 1080p might suck at 1440p because 8GB of VRAM is insufficient. Other games might favor larger memory caches, more memory bandwidth, etc. With super sampling technologies in play, image quality is not quantified in benchmark results. Newer technologies like DLSS 3 Frame Generation might smooth out image playback but introduces input latency. Et cetera ad nauseam.

After all, it's not like I play one game or visit one website.
A browser benchmark is not quite the same as benchmarking a single game. What you see is that there are hundreds of thousands of websites that run better on the WebKit engine, hundreds of thousands of websites run better on the Blink engine and hundreds of thousands of websites run better on the Gecko engine.

Of game benchmarks, you can say that for 1080p users they are largely irrelevant at the moment. What you see at 1080p in new games is that the RX 6600 reaches more than 80 fps in +- 75% of the new games at the highest settings. >95% of people can't actually tell the difference between 80 fps and 800 fps so you can say that going higher than 80 fps makes no sense.

So let's say the RX 6600 fails to reach 80fps in 25% of new games. In those games, you can almost always just adjust a (limited) number of settings and you get 80 fps.

About the VRAM limitation, this is also largely something psychological that usually doesn't matter much. If you start looking at the comparisons of lowest settings to highest settings of modern games, you will see that in many modern games there are often no big differences. In the YouTube comparisons I've seen, people overwhelmingly agreed that new games often looked fine on the lowest settings. If you have a VRAM bottleneck you can simply usually lower your textures and this usually has extremely little effect on your gaming experience. You probably also remember that many people used to play on Xbox and PS3 with much inferior graphics, and that the games then were not always worse experiences than the latest games that have many times sharper graphics.

Furthermore, there are also gamers who continue to play games released around the year 2000 or earlier and play little or no other games. This is an older generation then, but not yet people who are super old. Those prefer to play old games out of nostalgia and for them, 'new' game benchmarks have no value.

Another thing I would say is that there are a number of games that are largely programmed in JavaScript. Angry Birds has become the first game to achieve 500 million downloads.
Guess which programming language this insanely popular game is programmed in?
More and more frequently, JavaScript is used to build popular games like Angry Birds, Bejeweled, and Polycraft.

Netflix relies heavily on JavaScript. It even hosts JavaScript Talks on YouTube where they state that “we share our learnings and developments in the world of JavaScript and frontend engineering with the larger JavaScript community.”

The Uber ride-sharing app has grown significantly since its humble beginnings in 2009. And the magic behind its user-friendly, interactive interface is JavaScript.

Offering 38 languages and nearly 100 courses, Duolingo is the language-learning tool for becoming proficient in Spanish, English, Japanese, Arabic, Latin, Hebrew, and many other languages. Anyone who has used Duolingo to learn languages or take proficiency tests knows how interactive and visually appealing the app is. With its signature owl who pops up to urge learners to keep persisting or cheers them on for correct answers, Duolingo makes good use of JavaScript, one of its foundational languages.

In 2015, Facebook released the now enormously popular JavaScript framework React Native, which facilitates the process of building native apps with React.
The professional networking app LinkedIn also uses JavaScript. In 2012, when it was already well-established, the business released an overhauled application that heavily relied on JavaScript.

The fact that browser performance evolve faster than the performance of a game is, in my opinion, another argument for saying that it is more useful to benchmark browsers, as there is often no up-to-date info available, whereas with games a benchmark remains valid for much longer.
 
Joined
Jun 21, 2021
Messages
2,901 (2.74/day)
System Name daily driver Mac mini M2 Pro
Processor Apple proprietary M2 Pro (6 p-cores, 4 e-cores)
Motherboard Apple proprietary
Cooling Apple proprietary
Memory Apple proprietary 16GB LPDDR5 unified memory
Video Card(s) Apple proprietary M2 Pro (16-core GPU)
Storage Apple proprietary onboard 512GB SSD + various external HDDs
Display(s) LG 27UL850W (4K@60Hz IPS)
Case Apple proprietary
Audio Device(s) Apple proprietary
Power Supply Apple proprietary
Mouse Apple Magic Trackpad 2
Keyboard Keychron K1 tenkeyless (Gateron Reds)
Software macOS Ventura 13.6 (with latest patches)
Benchmark Scores (My Windows daily driver is a Beelink Mini S12 Pro. I'm not interested in benchmarking.)
This is a perfect example of how inane this browser benchmark is.

No one is saying that Javascript isn't important. However it is not the only web technology that websites use. And yes, some of us have been using the World Wide Web since its inception, when JavaScript was called NetScript. In a moment of marketing brilliance, it was renamed JavaScript when Java became hip and cool back in the mid-Nineties.

Throwing in Angry Birds is an odd choice of evidence of JavaScript's importance. It became popular as mobile game app (on iOS first for that matter). It wasn't like anyone had a choice how to access it. You downloaded the app to your iPhone and played the game.

Sure, you can point out a few notable examples of which sites are heavy users of JavaScript. Therein lies the crux of the matter. Joe Consumer isn't going remember this. Ask Joe Consumer to list five web technologies and three example sites of each. I doubt if Joe Consumer could list a single one.

And most of this is simply for desktop browsing anyhow. On my iPhone and iPad, I don't have a real choice. Everything uses the WebKit rendering engine, even Chrome and Firefox.

Gee, Chase uses Technology X, so I should use Browser A. But BankOfAmerica uses Technology Z so I should use Browser C. And Amazon uses Technology Y, but also X and Z, which browser should I use? Okay, LinkedIn uses JavaScript. I log in 2-3 times a year on that website. Am I or Joe Consumer going to remember to use a WebKit-based browser to gain some performance benefit? ESPN versus Yahoo Sports? Twitch versus YouTube versus Instagram versus Snapchat versus TikTok? Uploading photos to Shutterfly or Snapfish?

Which browser should I use? So confusing. Maybe if I can find that discussion thread on TPU, I can set the record straight and know which browser to fire up. Oh, and which browser should I be using to read TechPowerUp? Are there different browsers I should be using for the main news site and reading the forums?

Ahahahahahahahahahaha!!!!!

I have Mac, Windows 11, Windows 10, iPhone, and iPad all at my disposal. There's also a Raspberry Pi 4 with Raspbian on it somewhere in my house. Does it make sense for me to hunt around a specific hardware/software combination for each website? No, that's silly, borderline OCD. Oh look, the newest benchmark shows Chrome running on the latest Samsung Android smartphone runs this site/benchmark well. So should I go and buy one of those just to for that site?

This is a great example of how utterly nonsensical this benchmark is from a real world standpoint and you've done a fantastic job illustrating how ridiculous it is to focus on one web browser technology. Browser benchmarking is borderline pedantic nonsense that is really only relevant to people who write web browsers or maybe a handful who code the world's most heavily used websites.

If I visit a site a handful of times a year, it really doesn't matter to me which browser runs the best on it. It's most time consuming and wasteful to do the research trying to figure out which browser is going to save me hundreds of milliseconds. I don't give a crap about which web browser is fastest to pay my annual vehicle registration fees to the California DMV. I just need a web browser that works.

Hell, in the time it would take for me to read the eight comments in this discussion thread, I could have paid my DMV fees and had a beer and caught ten minutes of Women's World Cup action. Even on the browser that didn't have the best performance.
 
Last edited:
Joined
Jul 15, 2022
Messages
718 (1.07/day)
And most of this is simply for desktop browsing anyhow. On my iPhone and iPad, I don't have a real choice. Everything uses the WebKit rendering engine, even Chrome and Firefox.
Most of Chrome's source code comes from Google's free and open-source software project Chromium, but Chrome is licensed as proprietary freeware. WebKit was the original rendering engine, but Google eventually forked it to create the Blink engine; all Chrome variants except iOS used Blink as of 2017.

Throwing in Angry Birds is an odd choice of evidence of JavaScript's importance. It became popular as mobile game app (on iOS first for that matter). It wasn't like anyone had a choice how to access it. You downloaded the app to your iPhone and played the game.
If you start asking most people what the most popular game ever is they are probably going to answer Minecraft, GTA V, Fortnite, Overwatch, The Witcher 3, PUBG or Pokemon. But none of these popular games have more than 500 million downloads. The most popular game ever was probably programmed in JavaScript.

WebAssembly is even better suited to achieving high performance in both games and apps. I think there is a tendency for more and more games and apps to work through a browser in the future. You also see this with Office 365 that you can use it through a browser.

If you play a WebAssembly game and Firefox achieves 15% higher performance in the game than Chromium, surely it's worth opening Firefox instead of Chromium? Otherwise, what is the point of overclocking where a lot of extra energy is consumed for +-9% higher performance and often less.

How to improve Gaming performance in Microsoft Edge

10 Best Gaming Browsers 2023 - Rigorous Themes

INTERNET GROWTH STATISTICS

48% of people under 42 spend more time socializing online than off

What post-pandemic slump? Online shopping is on the rise

E-commerce continues to grow in the EU

The Rise of Progressive Web Apps: A New Era in Website Development

Youth spend triple amount of time online than watching TV: Research

Over 38% Say Teenagers Spend More Than 8 Hours on Social Media Daily

We now spend more than eight hours a day consuming media

Whether I would have an RX 6600 or an RX 7900 XTX, for gaming at 1080p it would be the exact same experience in most games I play.
Whereas when I use the R5 3400G PRO to browse the internet I notice that websites are significantly more responsive and load faster than on my i3-3240.
For me personally, browser benchmarks are more important than gaming benchmarks.
 
Joined
Nov 26, 2021
Messages
1,372 (1.52/day)
Location
Mississauga, Canada
Processor Ryzen 7 5700X
Motherboard ASUS TUF Gaming X570-PRO (WiFi 6)
Cooling Noctua NH-C14S (two fans)
Memory 2x16GB DDR4 3200
Video Card(s) Reference Vega 64
Storage Intel 665p 1TB, WD Black SN850X 2TB, Crucial MX300 1TB SATA, Samsung 830 256 GB SATA
Display(s) Nixeus NX-EDG27, and Samsung S23A700
Case Fractal Design R5
Power Supply Seasonic PRIME TITANIUM 850W
Mouse Logitech
VR HMD Oculus Rift
Software Windows 11 Pro, and Ubuntu 20.04
Hardware: Ryzen 7 5700X, 32 GB ECC DDR4 3200 CL22, Reference Vega 64, Intel 665p 1 TB (Boot drive)
OS: Windows 11 22H2
Links to detailed scores in the Time column

BrowserVersionTime (ms)Confidence Interval (%)
Firefox115.0.256.29.5
Chrome115.0.5790.9863.52.4
Vivaldi6.1.3035.11168.66.8
Edge114.0.1823.8269.48.5

I'll post results with Ubuntu later.

EDIT: Ubuntu 22.04.2 LTS
Hardware: same as above except boot drive is now a Samsung 830 256 GB SATA SSD

BrowserVersionTime (ms)Confidence Interval (%)
Firefox106.0.575.27.7
Vivaldi6.1.3035.11186.56.1
 
Last edited:
Joined
Sep 23, 2022
Messages
976 (1.63/day)
System Name Windows
Processor 13900K
Motherboard Pro Z790-A WiFi
Cooling Noctua NH-D15s
Memory 32GB 6600 CL32
Video Card(s) RTX 4090
Display(s) MSI MAG401QR
Case Phanteks P600s
Power Supply Vertex GX-1000
Software Win 11 Pro
Benchmark Scores They suck.
Hardware: Intel i7-12700k + 64GB RAM @ 3600MHz CL18 DDR4 dual-channel + AMD RX 6900 XT 16GB + PNY CS3040 2TB
Software: Firefox 115 + Pop!_OS + Cinnamon

Score: 46.2ms

JS_Benchmark.jpg
 
Joined
Nov 25, 2019
Messages
803 (0.49/day)
Location
Taiwan
Processor i5-9600K
Motherboard Gigabyte Z390 Gaming X
Cooling Scythe Mugen 5S
Memory Micron Ballistix Sports LT 3000 8G*4
Video Card(s) EVGA 3070 XC3 Ultra Gaming
Storage Adata SX6000 Pro 512G, Kingston A2000 1T
Display(s) Gigabyte M32Q
Case Antec DF700 Flux
Audio Device(s) Edifier C3X
Power Supply Super Flower Leadex Gold 650W
Mouse Razer Basilisk V2
Keyboard Ducky ONE 2 Horizon
HP prodesk 400 I3-10100T 8GB DDR4 RAM @3200MHz
Google chrome 114.0.5735.199
1689908908392.png
 
Joined
Jul 15, 2022
Messages
718 (1.07/day)
Capture d’écran du 2023-07-21 10-47-44.png


Result: 92.5 ms

Hardware: Intel i3-3240 + 8GB RAM @1600MHz DDR3 dual-channel + NVIDIA GTX 650 1GB + EVO 850 500GB
Software: NixOS 23.11 & Nyxt browser 3.1.0

Although Nyxt is one of the least popular browsers and isn't fully finished, it scores higher than all the other Linux browsers in this benchmark.
Epiphany browser was +-15% slower in SunSpider than Nyxt and both use WebKitGTK.
 
Joined
Feb 21, 2008
Messages
104 (0.02/day)
System Name i9-13900k
Processor Intel Core i9 13900K
Motherboard Asus ProArt Z790 Creator WiFi
Cooling Corsair H150i Elite Capellix XT
Memory Corsair Dominator 64gb 5600MT/s DDR5 Dual Channel
Video Card(s) Sapphire Nitro+ RX 7900 XTX
Storage Samsung 980 PRO 1 & 2 TB
Display(s) LG 50" QNED TV 120hz
Case Corsair RGB Smart Case 5000x (white)
Audio Device(s) Fiio K5PRO USB DAC
Power Supply Corsair RM1200x Shift
Mouse Logitech MX Ergo Trackball
Keyboard Logitech K860
Software Windows 11 Pro
See signature below for system specs
Browser Google Chrome Canary 117.0.5922.2

sunspider1.0-1.8.2023.jpg
 
Last edited:
Joined
Jul 15, 2022
Messages
718 (1.07/day)
Screenshot from 2023-08-02 14-22-22.png

Result: 80.1 ms

Hardware: AMD R5 PRO 3400G + dual-channel DDR4 @2666MHz + integrated graphics
Software: Clear Linux + i3 wm + Firefox 115.0.3 + Mesa open-source GPU driver
 
Joined
Jul 15, 2022
Messages
718 (1.07/day)
2023-10-04-094712_1920x1080_scrot.png

score 96.5 ms

Hardware: Intel i3-3240 + 8GB RAM @1600MHz DDR3 dual-channel + NVIDIA GTX 650 1GB + EVO 850 500GB
Software: Firefox 118 + FreeBSD 13.2 + bspwm
 
Joined
Jul 15, 2022
Messages
718 (1.07/day)

Result: 38.7 ms
Software: Clear Linux -- Nyxt 3.9.0 -- GNOME Shell -- nouveau GPU driver
Hardware: Intel 12600KF -- Kingston dual-channel 6000 MHz CL40 -- GTX 650 1GB -- BIOSTAR B760MZ-E PRO -- Antec P6 -- Xilence XP550 -- ARCTIC i35 -- EVO 850 500GB
 
Joined
Jul 15, 2022
Messages
718 (1.07/day)

Result: 35.4 ms (this is not my best result, the benchmark also scored 34.x ms several times)
Software: ROSA Fresh Desktop 12.4 -- LXQt -- Nvidia proprietary driver -- XFS as root -- Nyxt 3.10.0
Hardware: Intel 12600KF (stock) -- Kingston 6000 MHz CL40 -- GTX 650 1GB -- BIOSTAR B760MZ-E PRO -- Antec P6 -- Xilence XP550 -- ARCTIC i35 -- EVO 850 500GB
 
Joined
Aug 2, 2016
Messages
44 (0.02/day)
My little ryzen 7600 tuned system managed to reach 34ms. Maybe should try this benches on new build of catchyOS compiled with native x86v4 (avx) instruction set.
windows 1123h2 chrome 120

2.png
 
Joined
Nov 22, 2020
Messages
60 (0.05/day)
Processor Ryzen 5 3600
Motherboard ASRock X470 Taichi
Cooling Scythe Kotetsu Mark II
Memory G.SKILL 32GB DDR4 3200 CL16
Video Card(s) EVGA GeForce RTX 3070 FTW3 Ultra (1980 MHz / 0.968 V)
Display(s) Dell P2715Q; BenQ EX3501R; Panasonic TC-P55S60
Case Fractal Design Define R5
Audio Device(s) Sennheiser HD580; 64 Audio 1964-Q
Power Supply Seasonic SSR-650TR
Mouse Logitech G700s; Logitech G903
Keyboard Cooler Master QuickFire TK; Kinesis Advantage
VR HMD Quest 2
SunSpider is a poor benchmark according to engineers on both the V8 and SpiderMonkey development teams, which are the JavaScript engines in Chrome and Firefox respectively. Some quotes:
Speaking from the perspective of a (former) V8 engineer, SunSpider is crappy benchmark, and I can't believe people still track it. It has misdirected JavaScript performance engineering work for going on 15 years now, leading to all kinds of weird contortions in JavaScript engines that do not benefit real world sites.
We care about Speedometer, because it correlates reasonably well with real-world performance. We pick and choose which parts of Jetstream2 we bother with. Anything older than that is strictly a measuring stick, rather than a target.
Source: https://news.ycombinator.com/item?id=37134092

They are speaking from the perspective of browser benchmarking rather than CPU benchmarking. But given how sensitive this benchmark would be to differences in browser versions I would not bother using it for CPU's either.
 
Joined
Jul 15, 2022
Messages
718 (1.07/day)
SunSpider is a poor benchmark according to engineers on both the V8 and SpiderMonkey development teams, which are the JavaScript engines in Chrome and Firefox respectively.

The old JS functionality has not changed much over the past decades. According to tests, the skills of the average Apple programmer have declined over the last decade. Which means that the programmers who developed SunSpider were normally more knowledgeable than the programmers who developed speedometer 2.1.

Today, I measured the load times of the very most popular web pages.

Where Chromium is faster:
- YouTube
- Facebook

Where Firefox is faster:
- TikTok
- Wikipedia
- Instagram
- Yahoo

Where they are just as fast:
- Amazon

According to speedometer, Chrome is significantly faster than Firefox, which we do not see in load times of the most popular websites. Which means SunSpider is a more realistic benchmark than speedometer.

If you look at which of the two benchmarks correlates best with reality, it is SunSpider 1.0, and not speedometer 2.1


Result: 34.4 ms
Software: ROSA Fresh Desktop 12.4 -- LXQt -- Nvidia proprietary driver -- XFS as root -- Nyxt 3.10.0
Hardware: Intel 12600KF (stock) -- Kingston 6000 MHz CL40 -- GTX 650 1GB -- BIOSTAR B760MZ-E PRO -- Antec P6 -- Xilence XP550 -- ARCTIC i35 -- EVO 850 500GB
I did not do RAM tuning. You can also use RAM that is 37% faster if you would use a better motherboard. And you can also overclock.

The idle power consumption of the Intel 12600KF (stock) is also better than that of an overclocked Ryzen 7600 so Intel is apparently still the most efficient solution for office computers or many other types of use cases where the PC is often in idle.


The power consumption of the Intel 12600KF at idle is lower than that of the 12600K.
 
Joined
Aug 2, 2016
Messages
44 (0.02/day)
The idle power consumption of the Intel 12600KF (stock) is also better than that of an overclocked Ryzen 7600 so Intel is apparently still the most efficient solution for office computers or many other types of use cases where the PC is often in idle.
Efficient at idle? that also depends what kind of idle exactly. Properly configured b650system is idling at about 50watts at wall with igpu and dgpu attached. 40 watts without gpu, very similar to intel. Differences beetween them at idle has shrunk dramatically since 3-4 gens ago and now are comparable when looking apples to apples. There are a lot of shallow reviews in that regard, similar to those jedec or even expo ddr speeds. For example many gamer mobos at stock settings have turned on provision for powering up to 360 adressable leds. I had personally managed to bring down idle 10watts by lowering down that shit.

On my ryzen 7600 HwInfo is reporting power consumption of cores at just 13watts when idle on fixed 5.5ghz, on the other hand when idle at 3.8 on stock pbo settings it is just 2.5Watts idle. Both of those with overclocked ddr5 so it can go even lower. But 12400 is base clock only lowly 2.5ghz and on your screen at 1watt. 1.5 watt differences for much higher 3.8 base clock and snappiness. Something for something and whole thing is nothingburger.


But on a whole i agree with something. Every desktop pc idle power is joke, doing same thing as mobile for 10-20 x power. Intel dun gufed by not merging atx 12VO standard with atx 3.0. Their user buy new mobo every two gen anyway so i dont know why such pushback from press and "enthusiasts" when they were forced to buy new psu for that space heaters anyway. 12Vonly on desktop is much needed change, idle results were spectacular on early psus. Simpler mobos, lower power spikes, simple psu , and lots of power savings for the user for one time change. I hope EU will drag pc industry to finally doing it if california laws was not enogh.
 
Joined
Nov 22, 2020
Messages
60 (0.05/day)
Processor Ryzen 5 3600
Motherboard ASRock X470 Taichi
Cooling Scythe Kotetsu Mark II
Memory G.SKILL 32GB DDR4 3200 CL16
Video Card(s) EVGA GeForce RTX 3070 FTW3 Ultra (1980 MHz / 0.968 V)
Display(s) Dell P2715Q; BenQ EX3501R; Panasonic TC-P55S60
Case Fractal Design Define R5
Audio Device(s) Sennheiser HD580; 64 Audio 1964-Q
Power Supply Seasonic SSR-650TR
Mouse Logitech G700s; Logitech G903
Keyboard Cooler Master QuickFire TK; Kinesis Advantage
VR HMD Quest 2
Efficient at idle? that also depends what kind of idle exactly. Properly configured b650system is idling at about 50watts at wall with igpu and dgpu attached. 40 watts without gpu, very similar to intel. Differences beetween them at idle has shrunk dramatically since 3-4 gens ago and now are comparable when looking apples to apples. There are a lot of shallow reviews in that regard, similar to those jedec or even expo ddr speeds. For example many gamer mobos at stock settings have turned on provision for powering up to 360 adressable leds. I had personally managed to bring down idle 10watts by lowering down that shit.

On my ryzen 7600 HwInfo is reporting power consumption of cores at just 13watts when idle on fixed 5.5ghz, on the other hand when idle at 3.8 on stock pbo settings it is just 2.5Watts idle. Both of those with overclocked ddr5 so it can go even lower. But 12400 is base clock only lowly 2.5ghz and on your screen at 1watt. 1.5 watt differences for much higher 3.8 base clock and snappiness. Something for something and whole thing is nothingburger.
Okay, I'm veering way off topic now but I would really like to know more about what you did here. My system (see full spec in profile) idles no lower than 75-80W at the wall meter, and that's taking all reasonable measures in UEFI and Windows power settings to lower it.

UEFI stuff I tried: deactivating all unused devices; maximizing available P-states, C-states, and PCI-E sub-states. None seemed to change idle consumption significantly.

Win10 stuff I tried: maximizing all power savings features in "Ryzen Power Saver" power profile. The only settings with a measurable effect were spinning down HDDs (3 of 4 are usually spun down) and lowering max CPU clock speed. Capping CPU to 99% drops the processor's reported status from 4.15GHz/1.43V to 3.49GHz/1.06V and saves 5-10W at the wall when lightly loaded.

I typically run on "Ryzen Balanced" with max CPU at 99% and PCI-E optimizations off. I toggle to "Ryzen High Performance" for gaming.

"Ryzen Power Saver" saves another 5W at the wall (even though only 0.5W difference in HWInfo CPU PPT) but causes instability: I experienced several BSOD's over a week's time when PCI-E link state was set to maximum/L1. Dropping RAM from 3200 XMP to 2133 JEDEC saves 5+ watts at the wall, and lowers HWInfo SoC 3W, but tanks performance so I don't do that either. The RTX 3070 has idle fan stop and MSI AB reports 11-15W idle depending on background applications.

What's your Package Power / PPT value in HWInfo? Mine says my lightly-loaded Ryzen 3600 uses 21-22W package power / PPT. Breakdown is 2-3W core power, 9.8W SoC, and the rest unaccounted for.

I was going to go with Intel for my next upgrade because I heard total system power is less when lightly loaded. It is difficult to find precise comparisons between platforms online. Most reviewers don't report it at all, and those that do are wildly inconsistent with each other and publish no notes or methodology. Best info I could find was this site which doesn't do controlled comparisons but does have extensive notes on settings and hardware used.
 
Top