Discuss Scratch
- Discussion Forums
- » Advanced Topics
- » Mac
- b1048546
- Scratcher
1000+ posts
Mac
Yes. Regardless of their weird M1 graphs, it did perform extremely well. At least they named the graph.
- b1048546
- Scratcher
1000+ posts
Mac
I stand corrected! So no, in raw benchmarks, M1X doesn't beat the 105 watt RTX 3080 Mobile. In most real world tasks, the Max will win though.
Impressive performance, especially when it doesn't need to be plugged in to get it!
GFX Bench M1X (32 core GPU) beats the RTX 3080 Mobile in nearly every GFX Bench Gaming test. Onscreen and even the high-end tests. Fairly impressive. I suppose the OpenCL score doesn't mean much. Be a LITTLE cautious, as this test was really meant for smartphones. Although iPhones are on the same IP/architecture as M1 Max so I think it's a fairly accurate assumption. Although, I wouldn't be surprised if game developers were interested in macOS for once. (come on apple, give them more tools to convert.)
- MagicCrayon9342
- Scratcher
1000+ posts
Mac
meh, just port all the iOS games. there, thats what ya get mac ppl.I stand corrected! So no, in raw benchmarks, M1X doesn't beat the 105 watt RTX 3080 Mobile. In most real world tasks, the Max will win though.
Impressive performance, especially when it doesn't need to be plugged in to get it!
GFX Bench M1X (32 core GPU) beats the RTX 3080 Mobile in nearly every GFX Bench Gaming test. Onscreen and even the high-end tests. Fairly impressive. I suppose the OpenCL score doesn't mean much. Be a LITTLE cautious, as this test was really meant for smartphones. Although iPhones are on the same IP/architecture as M1 Max so I think it's a fairly accurate assumption. Although, I wouldn't be surprised if game developers were interested in macOS for once. (come on apple, give them more tools to convert.)
- b1048546
- Scratcher
1000+ posts
Mac
Except that they all already run on M1.meh, just port all the iOS games. there, thats what ya get mac ppl.I stand corrected! So no, in raw benchmarks, M1X doesn't beat the 105 watt RTX 3080 Mobile. In most real world tasks, the Max will win though.
Impressive performance, especially when it doesn't need to be plugged in to get it!
GFX Bench M1X (32 core GPU) beats the RTX 3080 Mobile in nearly every GFX Bench Gaming test. Onscreen and even the high-end tests. Fairly impressive. I suppose the OpenCL score doesn't mean much. Be a LITTLE cautious, as this test was really meant for smartphones. Although iPhones are on the same IP/architecture as M1 Max so I think it's a fairly accurate assumption. Although, I wouldn't be surprised if game developers were interested in macOS for once. (come on apple, give them more tools to convert.)
Last edited by b1048546 (Oct. 21, 2021 02:12:29)
- b1048546
- Scratcher
1000+ posts
Mac
Memory bandwidth on M1 Pro/Max is insane. I REALLY don't expect any benchmarks to show us anything, I only see real world tests working out for these chips.
My reasoning: I wouldn't even expect Cinebench R23 to make use of the 60 GB/s that one single FireStorm core uses. We actually know it can't. If it did, then the multi-core wouldn't have scaled well due to the M1 only having 69 GB/s of bandwidth. Few benchmarks and real-world tasks need this much bandwidth. Real-world performance will be much better due to 16 channels on the M1 Pro and 32 channels on the M1 Max. Although, Geekbench isn't the best benchmark so I'm wondering why the performance dropped a bit more than expected…
Side note: Intel and AMD literally don't need to switch to RISC/ARM based chips to compete. That's a LIE. For example, Alder Lake literally has the same big.little architecture as M1.
My reasoning: I wouldn't even expect Cinebench R23 to make use of the 60 GB/s that one single FireStorm core uses. We actually know it can't. If it did, then the multi-core wouldn't have scaled well due to the M1 only having 69 GB/s of bandwidth. Few benchmarks and real-world tasks need this much bandwidth. Real-world performance will be much better due to 16 channels on the M1 Pro and 32 channels on the M1 Max. Although, Geekbench isn't the best benchmark so I'm wondering why the performance dropped a bit more than expected…
Side note: Intel and AMD literally don't need to switch to RISC/ARM based chips to compete. That's a LIE. For example, Alder Lake literally has the same big.little architecture as M1.
- gdpr5b78aa4361827f5c2a08d700
- Scratcher
1000+ posts
Mac
16 gb of ram over 8 is a very big difference.I wouldn't count the ram being a benefit. The SSD in the Air is probably faster which is annoying because nobody asked for the fastest SSD lol. What about battery and screen? Trackpad, speakers, and keyboard? Would consider that to be more important for the general audience who's using one of these. man… an HP laptop with double the RAM and double the SSD capacity AND a better CPU. Same price as a Macbook Air. oof
- gdpr5b78aa4361827f5c2a08d700
- Scratcher
1000+ posts
Mac
of course the m1 is faster than a chip in the radiator computer. as far as i can see, unified memory isn't a good thing.
i'm not saying the gt 710 is faster than the m1 but people should stop “falling for” this marketing.
that's vs power no? an integrated gpu from a company known for making disposable earphones isn't going to beat nvidia's flagship desktop gpu.As per Apple, it does. M1 Max can't beat it
most of those specifications are just marketing nonsense. Can we just look at that power
i'm not saying the gt 710 is faster than the m1 but people should stop “falling for” this marketing.
- b1048546
- Scratcher
1000+ posts
Mac
That's true. As far as I'm aware, the LPDDR4X in the M1 is faster than the memory in the HP. The data doesn't need to be copied from the CPU or to the GPU on the M1 though. There's benefits to both for sure.16 gb of ram over 8 is a very big difference.I wouldn't count the ram being a benefit. The SSD in the Air is probably faster which is annoying because nobody asked for the fastest SSD lol. What about battery and screen? Trackpad, speakers, and keyboard? Would consider that to be more important for the general audience who's using one of these. man… an HP laptop with double the RAM and double the SSD capacity AND a better CPU. Same price as a Macbook Air. oof
Last edited by b1048546 (Oct. 21, 2021 16:43:40)
- b1048546
- Scratcher
1000+ posts
Mac
Yeah, I'm not going to believe it until real-world tests come out. (no, it's the laptop version btw) of course the m1 is faster than a chip in the radiator computer. as far as i can see, unified memory isn't a good thing.that's vs power no? an integrated gpu from a company known for making disposable earphones isn't going to beat nvidia's flagship desktop gpu.As per Apple, it does. M1 Max can't beat itmost of those specifications are just marketing nonsense. Can we just look at that power
i'm not saying the gt 710 is faster than the m1 but people should stop “falling for” this marketing.
The specs aren't marketing? Their comparisons may be fake, but the specs look normal.
- gdpr5b78aa4361827f5c2a08d700
- Scratcher
1000+ posts
Mac
the difference shouldn't be noticeable. That's true. As far as I'm aware, the LPDDR4X in the M1 is faster than the memory in the HP.
The specs aren't marketing? Their comparisons may be fake, but the specs look normal.yes but putting irrelevant specs in big purple caps can be a bit misleading
Last edited by gdpr5b78aa4361827f5c2a08d700 (Oct. 21, 2021 18:36:54)
- b1048546
- Scratcher
1000+ posts
Mac
True, less latency in some areas but in general it's a difference.the difference shouldn't be noticeable. That's true. As far as I'm aware, the LPDDR4X in the M1 is faster than the memory in the HP.
which ones?yes but putting irrelevant specs in big purple caps can be a bit misleading The specs aren't marketing? Their comparisons may be fake, but the specs look normal.
Last edited by b1048546 (Oct. 21, 2021 18:40:50)
- Chiroyce
- Scratcher
1000+ posts
Mac
while consuming less power it may do so - otherwise NVIDIA is obviously faster. an integrated gpu from a company known for making disposable earphones isn't going to beat nvidia's flagship desktop gpu.
April Fools' topics:
— New Buildings in Scratch's headquarters
— Give every Scratcher an M1 MacBook Air
— Scratch should let users edit other Scratchers' projects
— Make a statue for Jeffalo
— Scratch Tech Tips™
— Make a Chiroyce statue emoji
<img src=“x” onerror=“alert('XSS vulnerability discovered')”>
this is a test sentence
- Chiroyce
- Scratcher
1000+ posts
Mac
16 core Neural Engine is something they have and other laptops usually don't. ProRes encode and decode, again, is something HUGE for people who use ProRes. 64GB RAM and 400GBPS bandwidth are both accurate and both are going to have an impact. “industry leading performance per watt” is subjective, we'll have to see when YouTubers start getting the Macs. 5NM process, 4 displays, and CPU/GPU core counts are also accurate. Thunderbolt 4 and Secure Enclave was already there in M1. yes but putting irrelevant specs in big purple caps can be a bit misleading
Yeah - they could've at least mentioned the benchmarks and exact GPU + wattage they used. Yeah, I'm not going to believe it until real-world tests come out. (no, it's the laptop version btw)
Last edited by Chiroyce (Oct. 22, 2021 03:20:46)
April Fools' topics:
— New Buildings in Scratch's headquarters
— Give every Scratcher an M1 MacBook Air
— Scratch should let users edit other Scratchers' projects
— Make a statue for Jeffalo
— Scratch Tech Tips™
— Make a Chiroyce statue emoji
<img src=“x” onerror=“alert('XSS vulnerability discovered')”>
this is a test sentence
- b1048546
- Scratcher
1000+ posts
Mac
They did show the exact GPU and wattage they used. Performance per watt is better than anything AMD, Intel, or NVIDIA don't have the performance per watt that the M1 has. Apple's GPU IP is more impressive than their CPU IP.16 core Neural Engine is something they have and other laptops usually don't. ProRes encode and decode, again, is something HUGE for people who use ProRes. 64GB RAM and 400GBPS bandwidth are both accurate and both are going to have an impact. “industry leading performance per watt” is subjective, we'll have to see when YouTubers start getting the Macs. 5NM process, 4 displays, and CPU/GPU core counts are also accurate. Thunderbolt 4 and Secure Enclave was already there in M1. yes but putting irrelevant specs in big purple caps can be a bit misleadingYeah - they could've at least mentioned the benchmarks and exact GPU + wattage they used. Yeah, I'm not going to believe it until real-world tests come out. (no, it's the laptop version btw)
- Chiroyce
- Scratcher
1000+ posts
Mac
The part where the guy on the left speaks about the notch in this video, you can probably see that they already have the new Macs and have also used it. The date where the reviewers are allowed to drop their reviews are on the 25th. So can't wait!
Also it's been leaked by some guy lol - https://www.youtube.com/watch?v=k7Yiei5o3Ts
Also it's been leaked by some guy lol - https://www.youtube.com/watch?v=k7Yiei5o3Ts
April Fools' topics:
— New Buildings in Scratch's headquarters
— Give every Scratcher an M1 MacBook Air
— Scratch should let users edit other Scratchers' projects
— Make a statue for Jeffalo
— Scratch Tech Tips™
— Make a Chiroyce statue emoji
<img src=“x” onerror=“alert('XSS vulnerability discovered')”>
this is a test sentence
- b1048546
- Scratcher
1000+ posts
Mac
this video, you can probably see that they already have the new Macs and have also used it. The date where the reviewers are allowed to drop their reviews are on the 25th. So can't wait!Yeah, the date was already leaked lol. We have benchmarks though. The part where the guy on the left speaks about the notch in
Also it's been leaked by some guy lol - https://www.youtube.com/watch?v=k7Yiei5o3Ts
- b1048546
- Scratcher
1000+ posts
Mac
Still extremely skeptical about the GPU in M1 Pro and M1 Max.
CPU looks good though. (I'll be using Cinebench R23 as a reference because I don't like Geekbench all that much. Apple's M1 and M1 Pro/Max get the same Geekbench and Cinebench score.):
Single core is about the same as M1 but the multi-core (10-core) is ~12,500. That's really impressive. A 30 minute loop of Cinebench R23 gives you identical scores. With the 8-core CPU, it edges to about ~9,000 and it doesn't throttle. Obviously the 10-core is more impressive but this is pretty goodl.
CPU looks good though. (I'll be using Cinebench R23 as a reference because I don't like Geekbench all that much. Apple's M1 and M1 Pro/Max get the same Geekbench and Cinebench score.):
Single core is about the same as M1 but the multi-core (10-core) is ~12,500. That's really impressive. A 30 minute loop of Cinebench R23 gives you identical scores. With the 8-core CPU, it edges to about ~9,000 and it doesn't throttle. Obviously the 10-core is more impressive but this is pretty goodl.
- Chiroyce
- Scratcher
1000+ posts
Mac
Looks like Intel now took Apple's optimization strategy of dividing tasks into Performance and Efficiency cores - https://youtu.be/-lcruZQyh94
April Fools' topics:
— New Buildings in Scratch's headquarters
— Give every Scratcher an M1 MacBook Air
— Scratch should let users edit other Scratchers' projects
— Make a statue for Jeffalo
— Scratch Tech Tips™
— Make a Chiroyce statue emoji
<img src=“x” onerror=“alert('XSS vulnerability discovered')”>
this is a test sentence
- Chiroyce
- Scratcher
1000+ posts
Mac
JUST LOOK AT THIS https://youtu.be/uC6S9ywDSQg
April Fools' topics:
— New Buildings in Scratch's headquarters
— Give every Scratcher an M1 MacBook Air
— Scratch should let users edit other Scratchers' projects
— Make a statue for Jeffalo
— Scratch Tech Tips™
— Make a Chiroyce statue emoji
<img src=“x” onerror=“alert('XSS vulnerability discovered')”>
this is a test sentence
- Capitan_Florida_2
- Scratcher
100+ posts
Mac
https://youtu.be/-lcruZQyh94finally, they can rival phone chipsets in optimization Looks like Intel now took Apple's optimization strategy of dividing tasks into Performance and Efficiency cores -
cf2
being since 2016 __
I said “Nothing's new, nothing's changed, I still need you”.
note for me: CTRL+SHIFT+DOWN
i am very.
we just gotta get back to the website