Question Intel Raptor Lake vs AMD Zen 4 vs Apple M2

[Carfax83]

These CPUs are all going to square off against each other at some point this year assuming nothing catastrophic occurs to delay any of the product launches. So going by what we know from official sources and informed rumor mongers (many of which were very accurate before Alder Lake and the M1 launched), which CPU do you think will win out in these categories?

1) Single threaded performance
2) Multithreaded performance
3) Gaming performance
4) Performance per watt
5) Overall performance (who wins the majority of applications)

While I've been keeping a close eye on rumors and leaks for Zen 4 and Raptor Lake, I have not admittedly been doing so for the M2; as I'm unrepentant Apple hater  At least I'm honest about it... That said, this is my ranking based on what I've seen and heard:

I think the single threaded crown will go to Raptor Lake, and I say this based on informed rumors that Raptor Lake will have up to 10% more IPC from microarchitectural updates, cache upgrades and higher clock speeds than Alder Lake. From what I've seen, gauging IPC performance isn't easy as it varies so much based on application, but I'd say Alder Lake already has at least a 15% across the board IPC advantage over Zen 3, so Raptor Lake could conceivably have 25% better IPC than Zen 3, which is similar to what Zen 4 will reportedly possess. But I doubt Zen 4 will match Raptor Lake in clock speeds and memory latency performance, which is why I'm predicting Raptor Lake will take the single threaded performance crown.

For multithreaded performance, Zen 4 should easily take it due to having more big cores than its Intel counterpart and similar IPC.

Gaming performance is more complicated because while some games are inherently more reliant on single core performance (strategy games for instance), more and more 3D engines are becoming increasingly parallel due to the adoption of Vulkan and DX12 in addition to modernized programming methods. Still, very few 3D engines can scale beyond 8 threads and 6 to 8 cores remains the sweet spot for gaming and will be for some time. So overall, I feel more comfortable going with Raptor Lake for the gaming crown. Also if rumors are correct, Raptor Lake will officially support DDR5-5600 off the bat while Zen 4 will reportedly use DDR5-5200. The raw memory speed won't likely be a significant factor, but Intel's memory controller will be right next to the CPU cores while Zen 4's will be in the I/O die which while still on the same package will definitely incur a significant latency penalty; which I'm sure will be offset by a massive L3 cache.

On performance per watt, one would think the M2 should take this category easily......but from the small amount of research that I've collected on it, it seems that there won't be much of a performance increase with the M2, if any. Some rumors are even suggesting there may be a bit of a regression in that aspect. Also since Zen 4 will be on TSMC's 5nm node, it will undoubtedly have excellent performance per watt and I believe it will also easily crush Apple's best in single core. So for performance per watt, I'm going to go with Zen 4.

When it comes to overall performance, I'm leaning towards Zen 4 but it will be close. Raptor Lake will supposedly double the amount of Gracemont efficiency cores which will certainly help in multithreaded performance per watt, but ultimately they won't be a match for Zen 4's 16 big cores with SMT. AMD will have the core count advantage and when that's combined with IPC parity with Raptor Lake, Zen 4 will win the majority of the benchmarks.

 

[JoeRambo]

Not sure why you're bragging on the 12900k's compiler performance. 12900k on the Linux kernel compilation bench takes 200W vs 5950X 142W and performance is a dead heat. GNU Debugger compilation was almost as bad.

Average is 120W vs 144W. Nothing to write home about really. While beating 5950x in plenty of light threaded workloads in both efficiency and performance.
So even handicapped by Intel in default configuration it is performing real well.

 

[moinmoin]

I don't know man, I think Intel 7 is more power efficient than what most people think.

Go on the Intel Reddit board and you will find plenty of reports of people undervolting their 12900Ks and shaving off 50 to 80w off the stock power usage and exceeding the 5950x power efficiency in Cinebench.

Alder Lake is just horribly overvolted because Intel wanted to make a great showing in multithreaded performance vs the 5950x and 5900x.

It's sad that even in mobile Intel prefers to overvolt ADL. Waiting for ADL-U...

 

[insertcarehere]

It's sad that even in mobile Intel prefers to overvolt ADL. Waiting for ADL-U...

A dirty secret is that mobile TDPs have been fiction with both Intel and AMD for quite a while now.

Reality is, since laptop OEMs have had configurable power limits they have more often than not cranked them high to get maximum benchmark results.

https://www.notebookcheck.net/Lenov...erhouse-with-good-16-10-display.545205.0.html AMD 5900HX w/ 88w PL1 + PL2
https://www.notebookcheck.net/Asus-...etal-7-nm-AMD-Zen-3-Is-Stunning.540483.0.html AMD 5900HX w/ 80w PL1 + PL2
https://www.notebookcheck.net/Lenov...ming-laptop-with-a-lot-of-power.559368.0.html
AMD 5800H w/ 75w PL1/80w PL2.

For performance/gaming laptops, the default options more often than not will come "overvolted" out of the box regardless of CPU.

 

[MadRat]

The whole e-core vs p-core architecture seems misplaced until Microsoft and other OS makers can schedule it properly. If the e-core had both an IPC advantage and a clock advantage at full power over p-cores it would make sense, but otherwise what is the point? Your load redirected to the e-core is basically intended to be junk cycles by definition, which is an oxymoron when it comes to processing. Your simpler cores should scale faster than your most complex cores, so the logic for having e-cores is bass ackwards. They should have been more akin to double-pumped APUs than junk collectors.

AMD is scaling up their core counts AND partnering with Microsoft to take advantage of these core counts, and doing it without demanding tinkering with how the scheduler prioritizes work. That sounds like a better strategy.

And then Apple has premium process for their architecture. So its comparing a Porsche to a Corvette. Both are fast but the premium process allows you to look nimble in comparison. If you have money and specific tastes, the Porsche is a real option. But most of us could only afford the Corvette if we want a brand new hotrod. Whether its an Intel or AMD Corvette doesn't really matter, but your Intel has a slight higher overall cost and relatively higher fuel cost to go with a slight advantage on the drag strip.

 

[Abwx]

Well one of those two must clearly be wrong...

Intel Core® i9-12900K™ Review: A New Breed of CPU Flagship - News and Reviews

It’s no secret that Intel has remain rather stagnant in the last few years letting AMD to creep up in the CPU market. The Intel 12th Generation CPUs marks

hungrygeeks.com.ph

Here is another with similar scores.

Intel Core i9-12900K & Core i5-12600K im Test: Angriff mit 10 nm und Hybrid-Architektur

Alder Lake im Benchmark-Test: Wir haben die Leistungsfähigkeit in neun verschiedenen Anwendungen getestet und bilden daraus den Gesamt-Index.

www.pcgameshardware.de

On the graph you posted Zen 2 is faster than Zen 3 at same core count...

 

[TheELF]

On the graph you posted Zen 2 is faster than Zen 3 at same core count...

So?! Could be the difference in mobo or mem or just cooling. The rest of the numbers still roughly match up with the numbers from tech powerup except for the 12900k.

 

[eek2121]

The only way for AMD to achieve that would be to use 25-40% more transistors per core so that things that can use more instructions per cycle, like cinebench and so on, can use them and be faster, that would seriously cut into the amount of chips they can produce which would mean that their CPUs would have to be seriously more expensive to make up for it.

Boy do I have some news for you...

 

[eek2121]

A dirty secret is that mobile TDPs have been fiction with both Intel and AMD for quite a while now.

Reality is, since laptop OEMs have had configurable power limits they have more often than not cranked them high to get maximum benchmark results.

https://www.notebookcheck.net/Lenov...erhouse-with-good-16-10-display.545205.0.html AMD 5900HX w/ 88w PL1 + PL2
https://www.notebookcheck.net/Asus-...etal-7-nm-AMD-Zen-3-Is-Stunning.540483.0.html AMD 5900HX w/ 80w PL1 + PL2
https://www.notebookcheck.net/Lenov...ming-laptop-with-a-lot-of-power.559368.0.html
AMD 5800H w/ 75w PL1/80w PL2.

For performance/gaming laptops, the default options more often than not will come "overvolted" out of the box regardless of CPU.

TDP != power. That being said, AMD's chips generally consume a similar amount of power (plus extra for SoC). The 5900hx, on that page is likely configured for 54W. The SoC and related components use a bit of extra power that could potentially add up to 88w (I guess I can see that happening; I've never seen it in actuality). If you open up a software utility like HWInfo64, you can easily see how much power the CPU is using. My 5900hx, while running prime95 small FFTs, uses exactly 45W of power. This is while using the Geforce RTX 3070 GPU (vs. the onboard GPU) My laptop has software that lets me change the TDP on the fly, however, it does not ever exceed 54W even on max performance settings.

Intel on the other hand...

EDIT: Oh and for the record, total laptop power consumption during the prime95 test was 68W. That includes GPU, IO, etc.

 

[inf64]

I think that expecting RaptorLake to compete with Zen4 with Vcache in any workload is pure hopium. Even in ST Zen4 should win easily while running at lower clock.

 

[soresu]

I think that expecting RaptorLake to compete with Zen4 with Vcache in any workload is pure hopium. Even in ST Zen4 should win easily while running at lower clock.

Certainly where perf per watt is concerned unless they managed a drastic µArch power draw optimisation over Alder Lake.

 

[gruffi]

Well one of those two must clearly be wrong...

Here are similar results as TPU.

https://www.overclockers.com/intel-i9-12900k-and-i5-12600k-review/

Whatever. It's just an example. You should get the picture. You can find more apps with similar behavior. Where 12900K gets the best results with disabled E-core. Or where E-cores have only little impact on overall performance. Then it's still like 8 Intel cores vs 16 AMD cores. The same we have seen with previous generations. As long as Intel cannot offer more P-cores they are likely not winning the competition. E-cores can help them. But it's no sufficient answer to AMD's very power and area efficient Zen core.

 

[TheELF]

Here are similar results as TPU.

https://www.overclockers.com/intel-i9-12900k-and-i5-12600k-review/

Whatever. It's just an example. You should get the picture. You can find more apps with similar behavior. Where 12900K gets the best results with disabled E-core. Or where E-cores have only little impact on overall performance. Then it's still like 8 Intel cores vs 16 AMD cores. The same we have seen with previous generations. As long as Intel cannot offer more P-cores they are likely not winning the competition. E-cores can help them. But it's no sufficient answer to AMD's very power and area efficient Zen core.

We had similar benchmark results with threadrippers and "game mode" where performance would be cut in half, just because some software or some benchmarker uses some old or just wrong configuration doesn't mean that it can't be fixed.

The e-cores allow the p-cores to run at much higher clocks compared to the same CPU with only p-cores, for people that only use server type software it doesn't matter or might even be worse but for main stream users having 8 fast cores and some extra cores for all the background "junk" is much more beneficial.

 

[TheELF]

Boy do I have some news for you...

error 404...news not found.
If tsmc found a way to cheat physics and only told you then please do share.

 

[nicalandia]

I think that expecting RaptorLake to compete with Zen4 with Vcache in any workload is pure hopium. Even in ST Zen4 should win easily while running at lower clock.

Raptor Lake is just a refresh of Alder Lake in all fronts(Golden Cove has always been capable of 2MB L2$ and Garcemont core has always been capable of 4 MB L2$), don't expect more than 7% IPC at equal clocks. is just that the 13900K will have more MT Grunt due to additional 8 cores(e cores)

I expect the 13900K to do 36000 points in CBR23 MT,

I expect the 7950X(The 6000 line will be taken by Zen3+) to land between 34,000 to 36,000. in CBR23 MT. I just can't see it braking 40,000 points

 

[IntelUser2000]

Anandtech review concluded that the E cores have about half the performance of a P core in MT but 8P cores + HT is enough to almost reach the limits of TDP therefore the performance is limited.

I assume it's due to priotizing using the P cores first. So there is an optimization point they can use for Raptorlake, by reducing P core frequency by a tiny bit which will afford lot more power for the extra E cores.

 

[coercitiv]

error 404...news not found.

Here's the news homework for you:

• the leak we have so far puts Zen4 chiplet at roughly 70mm2 die area, smaller than ~80mm2 Zen3
• as @Rannar mentioned, 5nm scaling for logic is up to ~80%, so a 25%-35% absolute transistor count increase in a slightly smaller chiplet does not require TSMC to bend the laws of physics

Expect Zen4 to be a very balanced design, optimized for both good yields and performance increase. And also get used to the idea of Zen4 having better dynamic range than Zen3 did.

 

[Henry swagger]

Here's the news homework for you:

• the leak we have so far puts Zen4 chiplet at roughly 70mm2 die area, smaller than ~80mm2 Zen3
• as @Rannar mentioned, 5nm scaling for logic is up to ~80%, so a 25%-35% absolute transistor count increase in a slightly smaller chiplet does not require TSMC to bend the laws of physics

Expect Zen4 to be a very balanced design, optimized for both good yields and performance increase. And also get used to the idea of Zen4 having better dynamic range than Zen3 did.

We.ll wait for benchmarks to see if ipc claims are true

 

[Thunder 57]

...but for main stream users having 8 fast cores and some extra cores for all the background "junk" is much more beneficial.

LMFTFY

but for mainstream users whatever Intel decides to do is much more beneficial.

 

[Zucker2k]

AMD's power consumption measurements using software such as HWinfo is to be taken with "a grain of salt," quoting chipsandcheese. The reference article further states:

The results indicate that the energy data is modeled, not measured: Even workloads that do not use memory (sqrt, add_pd, mul_pd) show inconsistent power. Further, the energy consumption of memory accesses (e.g., memory_read, memory_write) is not fully captured by RAPL. No DRAM domain is available and the RAPL package domain reports significantly lower power compared to the external measurement. Considering the different domains, this does not necessarily imply that RAPL readouts are wrong. But it shows that they cannot be used to accurately estimate and therefore optimize for total system power, as opposed to Intel systems since Haswell. On such systems, this is possible when adding Package and DRAM energy [

Always wondered why system power consumption numbers see AMD consuming as much or even higher than Intel systems sometimes, especially in gaming, even when they're reporting lower fps, meaning the dgpu is doing less work? Whenever someone comes along and points this out (not just gaming), you hear defenders arguing system power is not core power. Well yes, but where is that extra power going? Intel boards come with equally packed features, if not more. AMD got away with reporting on-die temps as opposed to core temps, iirc in the phenom days or prior. I think this power reporting misreporting, or discrepancy between software reported power and wall measured power consumption needs some explanation. I'm sure someone will chime in with the IOD is the culprit. Well, given what's quoted above, it's not a simple matter of subtracting core power from system power, is it?

 

[itsmydamnation]

AMD's power consumption measurements using software such as HWinfo is to be taken with "a grain of salt," quoting chipsandcheese. The reference article further states:


Always wondered why system power consumption numbers see AMD consuming as much or even higher than Intel systems sometimes, especially in gaming, even when they're reporting lower fps, meaning the dgpu is doing less work? Whenever someone comes along and points this out (not just gaming), you hear defenders arguing system power is not core power. Well yes, but where is that extra power going? Intel boards come with equally packed features, if not more. AMD got away with reporting on-die temps as opposed to core temps, iirc in the phenom days or prior. I think this power reporting misreporting, or discrepancy between software reported power and wall measured power consumption needs some explanation. I'm sure someone will chime in with the IOD is the culprit. Well, given what's quoted above, it's not a simple matter of subtracting core power from system power, is it?

On both my Lenovo zen 2 and 3 laptops this doesn't match my own testing where hw info ryzen controller and wall all are basically the same. So a desktop "artifiact"?

 

[coercitiv]

Always wondered why system power consumption numbers see AMD consuming as much or even higher than Intel systems sometimes, especially in gaming, even when they're reporting lower fps

Here you go, enjoy the read.

TL;DR mobo makers did it for marketing purposes.

Explaining the AMD Ryzen "Power Reporting Deviation" -metric in HWiNFO

Ryzen CPUs for AM4 platform rely on external, motherboard sourced telemetry to determine their power consumption. The voltage, current and power telemetry is provided to the processor by the motherboard VRM controller through the AMD SVI2 interface. This information is consumed by the processors power management co-processor, that is responsible for adjusting the operating parameters of the CPU and ensuring, that neither the CPU SKU, platform or infrastructure specific limits are being violated.

The weakness of this method is, that the telemetry essentially uses an undefined scale for the current (and hence power) measurements. This means that the motherboard VRM controller will send an integer between 0 - 255 to the CPU, and based the reference value known by the co-processor firmwares, this integer is converted to a figure, that represents a physical current drawn by the CPU. Based on the accurately known current flow and the voltage, it is possible to calculate to CPU power draw in Watts (V * I).

The reference value mentioned earlier is generally different for each of the motherboard make and model, unless there are boards which have an identical power circuitry. Because of that, it is on the motherboard manufacturers responsibility to find the correct value for their motherboard design through the means of calibration, and then to declare it properly in AGESA, during the bios compile time. In case the motherboard design specific, correct value differs greatly from the declared value, there will be a bias in the power consumption seen by the CPU. In case the declared value is greater than the actual value, the power consumption seen by the CPU is greater than it actually is. Likewise, if the declared value would be an understatement... the CPU would think it consumes less power than it actually does.

Since at least two of the largest motherboard manufacturers, still insist on using this exploit to gain an advantage over their competitors despite being constantly asked and told not to, we thought it would be only fair to allow the consumers to see if their boards are doing something they're not supposed to do. The issue with using this exploit is, that it messes up the power management of the CPU and potentially also decreases its lifespan because it is running the CPU outside the spec, in some cases by a vast margin. Also, it can cause issues when this exploit goes undetected by a hardware reviewer, since both the performance and the sofware based power consumption figures will be affected by it.

 

[TheELF]

Here's the news homework for you:

• the leak we have so far puts Zen4 chiplet at roughly 70mm2 die area, smaller than ~80mm2 Zen3
• as @Rannar mentioned, 5nm scaling for logic is up to ~80%, so a 25%-35% absolute transistor count increase in a slightly smaller chiplet does not require TSMC to bend the laws of physics

Expect Zen4 to be a very balanced design, optimized for both good yields and performance increase. And also get used to the idea of Zen4 having better dynamic range than Zen3 did.

Yeah, so why are you comparing 7nm to 5nm? They are priced differently and one doesn't even cut into the production of the other.

Increasing the transistor count on 5nm will yield less cores per wafer than not increasing transistor count, there isn't even any room for discussion there, the die size you mention and link to doesn't even tell us if it's the same transistor count as previous zen which it very well could be.
TSMC already announced a 20% increase in all prices, if AMD increases that by another 25-35% by adding that amount of transistors more then the final price will be 45-55% higher which would be way too high.

 

[TheELF]

Here you go, enjoy the read.

TL;DR mobo makers did it for marketing purposes.

Explaining the AMD Ryzen "Power Reporting Deviation" -metric in HWiNFO

Yeah no, if you get on intel's case for allowing lifted power limits then this is AMD's equivalent of that, they could have redesigned it many times or had it designed right in the first place, just because AMD is all like don't do it we don't allow it...but "wink wink nudge nudge" doesn't make it right.

 

[eek2121]

AMD's power consumption measurements using software such as HWinfo is to be taken with "a grain of salt," quoting chipsandcheese. The reference article further states:


Always wondered why system power consumption numbers see AMD consuming as much or even higher than Intel systems sometimes, especially in gaming, even when they're reporting lower fps, meaning the dgpu is doing less work? Whenever someone comes along and points this out (not just gaming), you hear defenders arguing system power is not core power. Well yes, but where is that extra power going? Intel boards come with equally packed features, if not more. AMD got away with reporting on-die temps as opposed to core temps, iirc in the phenom days or prior. I think this power reporting misreporting, or discrepancy between software reported power and wall measured power consumption needs some explanation. I'm sure someone will chime in with the IOD is the culprit. Well, given what's quoted above, it's not a simple matter of subtracting core power from system power, is it?

Hence why I mentioned “from the wall” power as well.

 

[coercitiv]

Yeah, so why are you comparing 7nm to 5nm?

Sometimes I wonder if time flows on the same vector where you live. A major disturbance in the time-space continuum would explain many anomalies in your posts.