Solved! ARM Apple High-End CPU - Intel replacement

[Richie Rich]

There is a first rumor about Intel replacement in Apple products:

• ARM based high-end CPU
• 8 cores, no SMT
• IPC +30% over Cortex A77
• desktop performance (Core i7/Ryzen R7) with much lower power consumption
• introduction with new gen MacBook Air in mid 2020 (considering also MacBook PRO and iMac)
• massive AI accelerator

Source Coreteks:

 

[DrMrLordX]

It's a bigger number, so it must be better. Just like Windows 95 is better than Windows 10 . I've actually heard that be asked before.

Windows 95 had a better themesong. Actually I don't think Windows 10 has a themesong.

The framerates on that flight sim . . . oof.

 

[Richie Rich]

I think it speaks unhappyness from you guys because x86 CPU lost a lot of leadership in CPU industry:

• lost IPC lead since 2015 Apple A9
• lost architecture advantege since 2017 and Apple A11 with world's 1st 6xALU core
• lost overall perfomance lead since 2019 Apple A13 beats fastest Ryzen 3950X at 4.6 GHz
• lost easy frequency gains as x86 CPUs are stucked around 5 GHz
• lost majority money in IT business because smartphones become 7x bigger market that servers
• and will loose in SIMD width and performance next year thanks to ARMv9 and its 2048-bit SVE2

I don't understand why you put so much negativity and hate because something much better than x86 is coming. I wouldn't be worry, Santa Clause won't take your x86 PC back to North pole from you.

 

[DrMrLordX]

I think it speaks unhappyness from you guys because x86 CPU lost a lot of leadership in CPU industry:

No such thing has happened yet.

lost majority money in IT business because smartphones become 7x bigger market that servers

Nobody "lost money" because of smartphones. The cloud industry was made possible because of software vendors pushing functions off devices and onto remote servers. More smartphone sales = more server sales. You're daft if you don't see that.

I don't understand why you put so much negativity and hate

You seem confused. They aren't negative about ARM. They're negative about YOU. You routinely misrepresent Apple's products and ARM in general.

 

[Richie Rich]

They're negative about YOU. You routinely misrepresent Apple's products and ARM in general.

I see some of you breaking the forum rules to be anti-personnel focused.

I'm still waiting for your evidence that SPEC and Geekbench doesn't match reality. I'm still waiting for example of CPU which performs very well in SPEC but average real applications was much worse. Please stop excuses and give the evidence of your statement. Please show me how Zen, Zen2, SkyLake or Cortex delivers double IPC in SPEC than in real apps.

(Above portion is bolded by moderator)

You are not a moderator. If you believe someone is breaking a
forum posting guideline, you report the post.

AT Mod Usandthem

 

[DrMrLordX]

I see some of you breaking the forum rules to be anti-personnel focused.

I see someone being obnoxiously repetitive. What do you expect people to do? Debunking the same screed over and over becomes tiresome. You have done everything possible to make as many people on this forum hate Apple and ARM by being the "bad cheerleader".

I'm still waiting for your evidence that SPEC and Geekbench doesn't match reality.

As I have plainly stated, nobody reputable conducts a benchmark suite consisting only of Geekbench and SPEC. Nobody. Those are frequent subjects of leaks, but when actual CPUs launch, they run other software on it. Here is an example of what I would expect to see on a CPU to evaluate its performance relative to competing products:

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

Office

• Chromium Compile: Windows VC++ Compile of Chrome 56 (same as 2017)
• PCMark10: Primary data will be the overview results – subtest results will be in Bench
• 3DMark Physics: We test every physics sub-test for Bench, and report the major ones (new)
• GeekBench4: By request (new)
• SYSmark 2018: Recently released by BAPCo, currently automating it into our suite (new, when feasible)

System

• Application Load: Time to load GIMP 2.10.4 (new)
• FCAT: Time to process a 90 second ROTR 1440p recording (same as 2017)
• 3D Particle Movement: Particle distribution test (same as 2017) – we also have AVX2 and AVX512 versions of this, which may be added later
• Dolphin 5.0: Console emulation test (same as 2017)
• DigiCortex: Sea Slug Brain simulation (same as 2017)
• y-Cruncher v0.7.6: Pi calculation with optimized instruction sets for new CPUs (new)
• Agisoft Photoscan 1.3.3: 2D image to 3D modelling tool (updated)

Render

• Corona 1.3: Performance renderer for 3dsMax, Cinema4D (same as 2017)
• Blender 2.79b: Render of bmw27 on CPU (updated to 2.79b)
• LuxMark v3.1 C++ and OpenCL: Test of different rendering code paths (same as 2017)
• POV-Ray 3.7.1: Built-in benchmark (updated)
• CineBench R15: Older Cinema4D test, will likely remain in Bench (same as 2017)

Encoding

• 7-zip 1805: Built-in benchmark (updated to v1805)
• WinRAR 5.60b3: Compression test of directory with video and web files (updated to 5.60b3)
• AES Encryption: In-memory AES performance. Slightly older test. (same as 2017)
• Handbrake 1.1.0: Logitech C920 1080p60 input file, transcoded into three formats for streaming/storage:
  • 720p60, x264, 6000 kbps CBR, Fast, High Profile
  • 1080p60, x264, 3500 kbps CBR, Faster, Main Profile
  • 1080p60, HEVC, 3500 kbps VBR, Fast, 2-Pass Main Profile

Web

• WebXPRT3: The latest WebXPRT test (updated)
• WebXPRT15: Similar to 3, but slightly older. (same as 2017)
• Speedometer2: Javascript Framework test (new)
• Google Octane 2.0: Depreciated but popular web test (same as 2017)
• Mozilla Kraken 1.1: Depreciated but popular web test (same as 2017)

Legacy (same as 2017)

• 3DPM v1: Older version of 3DPM, very naïve code
• x264 HD 3.0: Older transcode benchmark
• Cinebench R11.5 and R10: Representative of different coding methodologies

I would expect the exact benchmark suite to be different in 2020. But you should have an idea of what users in this forum expect to see in a proper CPU benchmark. Hell even cell phone reviewers run more than just SPEC and GB5.

 

[Glo.]

I'm still waiting for your evidence that SPEC and Geekbench doesn't match reality.

Im still waiting for you to show Blender, Final Cut Pro X, Logic Pro X, KDENLive, etc Running on ARM CPUs faster than on x86.

 

[jpiniero]

Im still waiting for you to show Blender, Final Cut Pro X, Logic Pro X, KDENLive, etc Running on ARM CPUs faster than on x86.

That's what the upcoming WWDC is for, to kickstart app porting.

 

[Thala]

As I have plainly stated, nobody reputable conducts a benchmark suite consisting only of Geekbench and SPEC. Nobody. Those are frequent subjects of leaks, but when actual CPUs launch, they run other software on it. Here is an example of what I would expect to see on a CPU to evaluate its performance relative to competing products:

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

I would expect the exact benchmark suite to be different in 2020. But you should have an idea of what users in this forum expect to see in a proper CPU benchmark. Hell even cell phone reviewers run more than just SPEC and GB5.

Not sure what your point is. SPEC and Geekbench are a selection of individual benchmarks covering a wide range of tasks - which is including rendering, compression, compilation, simuilation etc..
Sure - one can always discuss if adding yet another compression benchmark like 7-zip to the list makes sense or not - but from my experience you will not get additional insights.

What is even more important is, that most benchmarks you suggest are only available as binary. They contain different code pathes with and without assembly, AVX, SSE optimizations etc. Compile flags are largely unknown. There is no guarantee that you really compare apples with apples.

 

[DrMrLordX]

Not sure what your point is. SPEC and Geekbench are a selection of individual benchmarks covering a wide range of tasks - which is including rendering, compression, compilation, simuilation etc..

I think I made my point pretty well already? Nobody reputable relies on those two tests alone to review a CPU.

 

[Markfw]

I think I made my point pretty well already? Nobody reputable relies on those two tests alone to review a CPU.

Unless you are an ARM supporter, and thats all there is, so everyone else has to live with it !

Except me. I am with you, more than ones that are small enough to fit in the cache. And more than single threaded.

 

[DrMrLordX]

Unless you are an ARM supporter, and thats all there is, so everyone else has to live with it !

Except me. I am with you, more than ones that are small enough to fit in the cache. And more than single threaded.

Well really, you CAN run a lot of benches on any ARM CPU that isn't from Apple. I could run a bunch of Linux command line benches on my phone (Snapdragon 855+). If I weren't so lazy. Might be fun though . . .

 

[Doug S]

Well really, you CAN run a lot of benches on any ARM CPU that isn't from Apple. I could run a bunch of Linux command line benches on my phone (Snapdragon 855+). If I weren't so lazy. Might be fun though . . .

The people whining about SPEC & Geekbench will have the same complaints there - Linux command line stuff isn't Windows which is what they really mean but won't admit it when they talk about "real applications".

 

[DrMrLordX]

Linux command line stuff isn't Windows

Go tell that to Phoronix? Those tests are pretty versatile (if perhaps leaning towards ST) and well-respected. Also plenty of respected open sources benchmarks are cross-platform.

 

[Thala]

I think I made my point pretty well already? Nobody reputable relies on those two tests alone to review a CPU.

With other words no argument from your side, except pointing at others.
And yes you can run lots of benches on ARM CPUs and they all paint the same picture as SPEC. You are complaining for the sake of complaining.
There is nothing magical in other benchmarks, that suddenly would change the picture.

 

[DrMrLordX]

With other words no argument from your side, except pointing at others.

I don't need other words. Go to any review site and read their articles as to why they choose the benchmark suite they choose. For example:

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

If that doesn't do it for you, I don't know what else to tell you. Nobody seems to want to address the idea that SPEC + Geekbench is "good enough for everyone" because the idea is considered to be completely moronic. Pardon me for adhering to best practices.

And yes you can run lots of benches on ARM CPUs and they all paint the same picture as SPEC.

Okey-dokey! So which part of SPEC is supposed to tell me how well a CPU will do in Blender? Honest question, I'm asking so you can tell me. Let's see what kind of picture it paints.

edit: I'm gonna go out on a limb and guess 526.blender_r ?

You are complaining for the sake of complaining.

There's this one forum poster here - not you - that likes to hold up A13 SPEC numbers and pretend like that's the end all, be all of data points. Intel is so far behind! AMD is so far behind! They lose in SPEC! IPC so low etc. It gets tiresome. So I asked him how well it did in other benchmarks. Crickets. Part of the reason was to point out that running other benchmarks that are commonly used on PCs on iOS is difficult if not impossible. Even if it's something open source that could be compiled. No matter how good is Apple's hardware, it's still locked up behind iOS (for now), making most of that particular poster's love of their hardware a bit silly.

 

[Thunder 57]

I don't need other words. Go to any review site and read their articles as to why they choose the benchmark suite they choose. For example:

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

If that doesn't do it for you, I don't know what else to tell you. Nobody seems to want to address the idea that SPEC + Geekbench is "good enough for everyone" because the idea is considered to be completely moronic. Pardon me for adhering to best practices.

Well put. What's the point of reviews if we can just run GB and SPEC and have all the answers?

There's this one forum poster here - not you - that likes to hold up A13 SPEC numbers and pretend like that's the end all, be all of data points. Intel is so far behind! AMD is so far behind! They lose in SPEC! IPC so low etc. It gets tiresome. So I asked him how well it did in other benchmarks. Crickets. Part of the reason was to point out that running other benchmarks that are commonly used on PCs on iOS is difficult if not impossible. Even if it's something open source that could be compiled. No matter how good is Apple's hardware, it's still locked up behind iOS (for now), making most of that particular poster's love of their hardware a bit silly.

Crickets lol. Maybe that guy finally came to reason? Or maybe he's just on vacation?

 

[Markfw]

I don't need other words. Go to any review site and read their articles as to why they choose the benchmark suite they choose. For example:

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

If that doesn't do it for you, I don't know what else to tell you. Nobody seems to want to address the idea that SPEC + Geekbench is "good enough for everyone" because the idea is considered to be completely moronic. Pardon me for adhering to best practices.



Okey-dokey! So which part of SPEC is supposed to tell me how well a CPU will do in Blender? Honest question, I'm asking so you can tell me. Let's see what kind of picture it paints.



There's this one forum poster here - not you - that likes to hold up A13 SPEC numbers and pretend like that's the end all, be all of data points. Intel is so far behind! AMD is so far behind! They lose in SPEC! IPC so low etc. It gets tiresome. So I asked him how well it did in other benchmarks. Crickets. Part of the reason was to point out that running other benchmarks that are commonly used on PCs on iOS is difficult if not impossible. Even if it's something open source that could be compiled. No matter how good is Apple's hardware, it's still locked up behind iOS (for now), making most of that particular poster's love of their hardware a bit silly.

I agree 100%. What nobody seems to get, or understand, is that Apple and ARM AT THE MOMENT seems to be very strong in single core, non IO dependent benchmarks. They were designed for that purpose and do it well. But what about things that have high IO and multi-threaded requirements ? Blender, is just one. What about a huge database server, that serves Anandtech ? or Amazon ? Or a medical database that serves 10 million consumers records, and is used everyday for 1000 different purposes ? I used to work for such a company. With data centers measured in square miles. They require terrabytes of memory, and the amount of IO for an Oracle database is staggering. The database is 100's of terrabytes in size, maybe thousands. Just my one little system had a 300 terrabyte database, and that is summarized data for analytical purposes, no details.

 

[DrMrLordX]

I agree 100%. What nobody seems to get, or understand, is that Apple and ARM AT THE MOMENT seems to be very strong in single core, non IO dependent benchmarks. They were designed for that purpose and do it well. But what about things that have high IO and multi-threaded requirements ? Blender, is just one. What about a huge database server, that serves Anandtech ? or Amazon ? Or a medical database that serves 10 million consumers records, and is used everyday for 1000 different purposes ? I used to work for such a company. With data centers measured in square miles. They require terrabytes of memory, and the amount of IO for an Oracle database is staggering. The database is 100's of terrabytes in size, maybe thousands. Just my one little system had a 300 terrabyte database, and that is summarized data for analytical purposes, no details.

ARM can be made to run Linux (or you can run Linux applications on Android through various semi-convoluted means). That opens up a lot of benchmark possibilities, assuming you can get adequate memory + storage to run the benchmark in question. That sort of eliminates huge database benchmarks.

A13, on the other hand . . .

Crickets lol. Maybe that guy finally came to reason? Or maybe he's just on vacation?

Last seen on May 3rd. Don't know anything else.

 

[Richie Rich]

With other words no argument from your side, except pointing at others.
And yes you can run lots of benches on ARM CPUs and they all paint the same picture as SPEC. You are complaining for the sake of complaining.
There is nothing magical in other benchmarks, that suddenly would change the picture.

You answer to people who don't want to understand. That's useless until somebody boot up Linux at A14 device and run some benchmarks like Blender. For example Blender results corresponds with GeekBench4 score pretty well:

Blender:

• Zen2 Ryzen 3700X ... 7463 s/GHz
• Cortex A72 (RPi4) ... 15443 s/GHz .... that's 48% of Zen2

GB4:

• Zen2 Ryzen 3700X ... 1291 pts/GHz
• Cortex A72 (RPi4) ...... 645 pts/GHz .... that's 50% of Zen2

 

[Hitman928]

You answer to people who don't want to understand. That's useless until somebody boot up Linux at A14 device and run some benchmarks like Blender. For example Blender results corresponds with GeekBench4 score pretty well:

Blender:

• Zen2 Ryzen 3700X ... 7463 s/GHz
• Cortex A72 (RPi4) ... 15443 s/GHz .... that's 48% of Zen2

GB4:

• Zen2 Ryzen 3700X ... 1291 pts/GHz
• Cortex A72 (RPi4) ...... 645 pts/GHz .... that's 50% of Zen2

Where are your blender numbers coming from?

 

[DrMrLordX]

A72? Oh good grief.

 

[Richie Rich]

Where are your blender numbers coming from?

My own measurements. RPi4 needs to run 64-bit Ubuntu unless it loose a lot of performance with original Raspbian (32-bit).

• 32-bit .... ST: 19813 s.... MT 5275 s
• 64-bit .....ST: 10295 s.... MT 4077 s

A72 benefits from 64-bit ISA by 29% in MT and incredible 92% in ST performance.

I bought RPi4 4GB just for fun knowing its super slow. But when I connect 4K 32 inch monitor to it's pretty usable office machine. As IPC being at half of Zen2 which with clock at 1.5GHz is equivalent to Zen2 at 750 MHz but it works pretty good with lightweight desktop. I can imagine that future RPi5 with 4-core A78 Hercules with IPC about Zen3 and clock at 2.5 GHz can be full replacement of cheap x86 PCs (Zen1, Zen2, Atom/Tremont cores, Skylake)... and for 50$. That'd be pretty amazing.

Yeah, but still it's funny to wait for Blender BMW render using 1-core for 5h and 30mins at RPi4

 

[soresu]

My own measurements. RPi4 needs to run 64-bit Ubuntu unless it loose a lot of performance with original Raspbian (32-bit).

• 32-bit .... ST: 19813 s.... MT 5275 s
• 64-bit .....ST: 10295 s.... MT 4077 s

A72 benefits from 64-bit ISA by 29% in MT and incredible 92% in ST performance.

Is that why my RPi4 is slow as balls?

Damn, kinda wish they would warn people about Raspbian, or pursue a better default OS package.

 

[Hitman928]

My own measurements. RPi4 needs to run 64-bit Ubuntu unless it loose a lot of performance with original Raspbian (32-bit).

• 32-bit .... ST: 19813 s.... MT 5275 s
• 64-bit .....ST: 10295 s.... MT 4077 s

A72 benefits from 64-bit ISA by 29% in MT and incredible 92% in ST performance.

I bought RPi4 4GB just for fun knowing its super slow. But when I connect 4K 32 inch monitor to it's pretty usable office machine. As IPC being at half of Zen2 which with clock at 1.5GHz is equivalent to Zen2 at 750 MHz but it works pretty good with lightweight desktop. I can imagine that future RPi5 with 4-core A78 Hercules with IPC about Zen3 and clock at 2.5 GHz can be full replacement of cheap x86 PCs (Zen1, Zen2, Atom/Tremont cores, Skylake)... and for 50$. That'd be pretty amazing.

Yeah, but still it's funny to wait for Blender BMW render using 1-core for 5h and 30mins at RPi4

No offense, but I'm going to ignore your Blender numbers, there's nothing here that says they are trustworthy. It's not personal, just factual. I would have the same reaction no matter who posted this.

 

[soresu]

I can imagine that future RPi5 with 4-core A78 Hercules

Not gonna happen.

Maybe if 8nm becomes ridiculously cheap as everyone else chases sub 5nm nodes - it's a case of expense in the end, there's a pretty good reason BCM2711 is made on 28nm instead of 16/14/12nm which many products like Fire TV device SoC's are being made on now.

The Pi Foundation guys have no other income than sales the Pi and its accessories to defray the costs of the SoC - unlike Amazon who have an entire empire, not to mention the content streamed to it from Amazon Video which is the real money maker for them.

Bear in mind that even A72 is having severe thermal issues without a heatsink in BCM2711 - A78 would probably not be viable even with 16/12nm.

At best I would expect A73, or at a stretch A75 in RPi5 as A73 is much more power efficient than A72, and A75 supposedly has a similar power/clock figure to A73.