News:

Willkommen im Notebookcheck.com Forum! Hier können sie über alle unsere Artikel und allgemein über Notebook relevante Dinge disuktieren. Viel Spass!

Main Menu

Asus ROG Strix Scar 17 G733PY laptop review: Intel is no longer king

Started by Redaktion, April 11, 2023, 22:55:17

Previous topic - Next topic

Redaktion

Asus' latest gaming laptop is one of the fastest in the market and it doesn't even need any Intel parts. You'll still want to keep a close eye on those high core temperatures, though.

https://www.notebookcheck.net/Asus-ROG-Strix-Scar-17-G733PY-laptop-review-Intel-is-no-longer-king.706291.0.html

RobertJasiek

This test measures 46+ dB for Witcher 3 while other reviews have found 40 ~ 62 dB during 3D gaming. Apparently, one can measure whatever one wants or, to put it mildly, tweak settings for very different impacts.

NikoB

Again EPIC FAIL from AMD (or Asus?) with a memory controller. The pumping speed does not even reach the infamous 60Gb / s, instead of 80+ for Raptor Lake controllers. And this despite the fact that even Intel controllers lose to the Apple M2 Max memory controller almost 5 times in memory pumping speed!

And against this absolutely shameful background, AMD has the audacity to announce 28 free pci-e 5.0 links in 7945HX! The total requirement is 28 x 3.938Gb/s = 110.2Gb/s at 100% load on the memory controller! 100%! But in reality, this load should not exceed 10-15% of the memory bandwidth!

Having on board a 7945HX with free 28 pci-e 5.0 links (a monstrous 56 pci-e 4.0 links), it's just a shame for a laptop of this level not to have at least 2 USB40, and preferably 2 more usb-a 3.2 gen 2x2 on the right.

With so many pci-e 5.0 links of their own, Asus, why isn't there 10Gbps RJ45?! A top-end laptop should have a top-end network, at least 5Gbps+.

By the way, owners of all 4k@144Hz monitors - do you know that all your monitors work only in DSC mode (lossy compression is similar to mp3 in sound), but not in lossless data transfer mode at 144Hz with HDR, 4:4:4? All of these monitors have HDMI 2.1 slyly clipped to 24Gbps. Thus, without the presence of TB4 (I'm not sure about USB40, because nowhere is it written that it supports 2xDP and video data tunneling in TB4 mode up to 40Gbps), none of you, the owners, will be able to get lossless 4k@144Hz with HDR 4 :4:4...in the "top" Asus laptop...

A screen with low contrast (in such a model it should be at least 1500:1) and brightness (that's why even a crutch - HDR400 - that is not declared).

Obvious laptop flaws - no 4k@120-144Hz IPS with 1500: 1+ (such panels have been on the market for a long time), webcam from the early 2000s - shameful 720p with fixed focus instead of adequate 1440p@60fps with autofocus for 2023.

An irrational empty space inside - either it was necessary to put a large battery or, similarly to the N550 series, place 4 additional speakers in front for powerful sound.

If it weren't for the poor quality screen and webcam, as well as the ridiculous savings on ports on the right (when there are so many extra pci-e 5.0 links in the processor), I would consider this model to buy, but Asus, again by, like Lenovo screwed up with Legion in 2023...

NikoB

In this review of another Asus model with 7945hx, the same results indicate either Asus' problems with configuring the 7945HX memory controller or architectural errors in the design of the AMD Zen4 memory controller:
notebookcheck.net/Asus-ROG-Strix-G17-G713PI-Gaming-laptop-impresses-in-the-test-with-the-new-Ryzen-9.701633.0.html

And in this review:
notebookcheck.net/Asus-ROG-Zephyrus-Duo-16-Notebook-Review-Multitasking-monster-with-AMD-Zen4-RTX-4090-Mini-LED.701152.0.html
already the 3rd model from Asus with 7945HX, there are no tests from AIDA64 on the memory controller at all, although they should be there, apparently the author of the review, having looked at the shameful results, decided not to even upload them, so as not to dishonor Asus and AMD...

Now compare the pumping speed in memory with the 13980HX controller in
notebookcheck.net/Alienware-m18-R1-laptop-review-Bigger-and-heavier-than-the-MSI-Titan-GT77.703874.0.html

Why is AMD so disgraced in its "top" processor with a memory controller, even against the backdrop of Intel?

riklaunim

Quote from: NikoB on April 12, 2023, 13:07:13Again EPIC FAIL from AMD (or Asus?) with a memory controller. The pumping speed does not even reach the infamous 60Gb / s, instead of 80+ for Raptor Lake controllers. And this despite the fact that even Intel controllers lose to the Apple M2 Max memory controller almost 5 times in memory pumping speed!

Why do you have to post this nonsense rambling under every review?


Apple has a unified memory design which is much different than your average DDR5 design and the throughput will be much higher for Apple. Yet in actual benchmarks or games/productivity, this throughput does not matter until you go to some edge-case scenario.


Quote from: NikoB on April 12, 2023, 13:07:13Why is AMD so disgraced in its "top" processor with a memory controller, even against the backdrop of Intel?

Dragon Range laptops will just get positive reviews with high CPU/GPU performance if designed well and they will sell to those that want such a device. There will be no "disgraced" or moaning that specific artificial AIDA64 benchmarks aren't "the best".

Blastard

puts "Intel is no longer king" in the title

..then proceeds to show intel's 13700HX winning every gaming benchmark at lower resolutions where the games are CPU bottlenecked, which is the one benchmark that the "king" of CPUs should be able to win..?

..then conveniently leaves out i9+4090 intel configs from the comparisons at high resolutions, instead comparing with 4070 and 4080..?

Nedim Tabakovic

In DOTA 2 my last year laptop beats it:

notebookcheck.net/Razer-Blade-17-laptop-review-RTX-3070-Ti-or-RTX-3080-Ti-Which-is-the-superior-Blade.618411.0.html

NikoB

Quote from: riklaunim on April 12, 2023, 13:40:55Apple has a unified memory design which is much different than your average DDR5 design and the throughput will be much higher for Apple. Yet in actual benchmarks or games/productivity, this throughput does not matter until you go to some edge-case scenario.

Dragon Range laptops will just get positive reviews with high CPU/GPU performance if designed well and they will sell to those that want such a device. There will be no "disgraced" or moaning that specific artificial AIDA64 benchmarks aren't "the best".
Think again about your stupid answer - I have clearly written how the bottleneck is DDR5 4800 in the shameful 128-bit mode for cpu with 28 pci-e 5.0 lanes.

And using the 7945HX for moronic games is like hammering nails with a microscope. But now, yes, such an era that most bipedal apes hammer nails with microscopes. After all, they are allowed to do this ...for the time being they are allowed.

PHVM_BR

"Our 2023 model is generally more power-hungry than the Intel-powered 2022 model. When running Prime95, for example, power consumption would stabilize at 205 W vs. just 117 W on last year's i9-12950HX configuration. Similarly, running Witcher 3 would consume about 24 percent more power over the older RTX 3080 Ti configuration. Performance may be faster gen over gen, but consumption has risen as well."

Allen,

This is a classic example where the absence of a test pattern interferes with the result, resulting in wrong information.

The Scar 17 SE (2022) was tested in Performance mode, where the notebook consumes considerably less than in Turbo mode, also delivering less performance...

riklaunim

Quote from: NikoB on April 12, 2023, 19:41:39I have clearly written how the bottleneck is DDR5 4800 in the shameful 128-bit mode for cpu with 28 pci-e 5.0 lanes.

An average laptop will have a dGPU x8 4.0, then 1-2 NVMe SSDs so x4 + x4 4.0 which ends up x8 5.0 of bandwidth. 20 leftover lanes will not be used, aside of maybe a Clevo 3-4kg brick with 4 SSDs and x16 GPU connection.

As for RAM "speed" - complain to JEDEC and laptop/CPU vendors that they don't use XMP in laptops and stick to low-ish JEDEC standards. 4800MT/s is most common, while 5200 and 5600 are newer, yet they aren't as good (frequency/timings) as desktop kits with XMP and option to set timings by hand.

And quad-channel RAM design would also help if you want a true DTR workstation. Especially when sticking to JEDEC SODIMMs.

And if you really need fast RAM in a laptop I'm sure vendors will please you with "premium" price tag for going over 8/16GB of LPDDR5X-8533 which will be soldered and that 32GB will be marked up similarly to Apple. 64GB unlikely. At that point it's just better to use a desktop than pretend a desktop AMD/Intel CPU put in a laptop can be a "DTR Workstation".


Quote from: NikoB on April 12, 2023, 19:41:39And using the 7945HX for moronic games is like hammering nails with a microscope. But now, yes, such an era that most bipedal apes hammer nails with microscopes.
If you want to use it for something else would imply you have the money or can be forced to spend money... so a workstation, even with same silicon but without limitations will cost you premium just because they can make you pay for it (like Nvidia did with Quattro cards or how Intel/AMD have "Pro" variants of their CPUs).

Oleksa

NikoB
I agree with you on many points. Except for the hatred of memory speed.  It's worth remembering that Apple is not limited to proprietary solutions. While others are.
Simple examples from the world of video cards and memory for them, as well as amd processors. Nvidia, having a faster GPU, faster memory speed, loses in the same test because AMD provided more memory. And that's not all, if you look not only at the graphics, but also at the image, nvidia also shows a very bad picture. AMD is at the bottom in the pictures, but in reality it is at the top (and for less money). Now let's take the problem of the 3rd generation ryzen on chips, where the write speed to memory was half as fast. But this did not stop AMD from being on top (with the exception of Intel with top solutions 1-2% ahead). So, again, AMD loses by numbers, but in reality, it is a winner.
Similarly, with Apple, the reality is different, they use a lot of RAM to reduce the soldered SSD, and most of the memory is used by the built-in accelerators. This is already a big chunk of performance. And in reality, m1-m2 are not as big as they are hyped, it also has to do with the OS (which is controversial, like other OSes).

Let's move on to laptops, when I was choosing a laptop, I looked at the numbers and saw Intel as the hero, the winner. But when you come to the store, in a few minutes you will find out how terrible Intel laptops are, they will not use that higher RAM speed, because in reality Intel will slow down due to temperature, dropping the frequency, and will not reach that paper speed.
We are waiting for further developments of ARM laptops for Windows|Linux, and then we will be able to specifically figure out who is who, and not praise the terrible slavery of Apple's eco-style.
We need to take a comprehensive solution and look at the real problem, not just benchmarks. The memory speed benchmark will not show the real result when I have 50% of the memory reserved and several dozen programs using this memory are running, and what will be the results in comparison. It's like with supercars: on paper, it will get from point A to point B faster than a regular sedan. But in reality, the sedan will get there, and the supercar will not, because it will run out of fuel and overheat the engine, with a high risk of getting into an accident.

I would like to say that AMD's problem is the number of mobile chips, but there is nothing wrong with the number, it's just that vendors use them not in laptops, but in hype consoles and mini-PCs. Also, a large chip budget goes to adult consoles. Therefore, this is not a problem for AMD, the chips appear on time, but not in laptops (the only problem is laptop manufacturers who cut back on functionality).

Mnu

Quote from: Blastard on April 12, 2023, 14:33:38puts "Intel is no longer king" in the title

..then proceeds to show intel's 13700HX winning every gaming benchmark at lower resolutions where the games are CPU bottlenecked, which is the one benchmark that the "king" of CPUs should be able to win..?

..then conveniently leaves out i9+4090 intel configs from the comparisons at high resolutions, instead comparing with 4070 and 4080..?

Haha, precisely.
However, for the people that play Cinebench R23, it is the king.
But for the people that actually play games on a gaming laptop, the R9 7945HX is not exactly the king...


AHighEndLaptopUser

Hi everyone,

just posting my thoughts here after having used this Laptop for more than half a year intensely. Both as a high-end 3D workstation and occasional gaming.

I bought this laptop, specifically because it is supposed to be less prone to throttling than Intel CPUs.

However, in heavy 3D workloads, the CPU throttles to 1GHz or less, very often.

I downloaded a tool called "G-Helper" to increase the fan speed and decrease the CPU voltage. That helped with temperatures and made throttling less common.

Still a disappointment.


Other Observations:
More USB ports should be present, 128 GB ram support would be great. Screen should be higher quality at this price point.

The software ASUS ships is buggy, fan curves don't work for example. G-Helper does a much better job.

More BIOS/UEFI options would be nice.

Better speakers are needed, too.

NikoB

Quote from: AHighEndLaptopUser on March 09, 2024, 05:37:41However, in heavy 3D workloads, the CPU throttles to 1GHz or less, very often.
This is the problem with local "reviews" - they are generally increasingly useless.

The authors test the processor separately and then the games, but do not indicate PL1 in this mode anywhere in the text, i.e. stable processor performance at full load on igpu/dgpu.

Therefore, local reviews from the point of view of buying a work tool rather than a toy machine are useless. They only care about gaming performance. And the speed of the processor almost always remains behind the scenes.

You will not find in the text of this "review" a single mention of PL1/PL2 in different modes and under complex loads.

This is not a professional review, but an amateur one. And all the reviews here look like that (or marketing ones) because there is no clear formalization and requirements for reviews. This is simply chaos and a mess, in which you simply cannot trust any authors or figures, because there are constant questions about their inconsistency with reality. This can be seen even when cross-comparing different reviews.

An ordinary amateur buyer, of course, will not notice this, but an experienced person with expert knowledge...


Therefore, never rely on one review when purchasing, read at least 3-4 independent ones and read 2-3 specialized forums on this model. Ask the owner questions. Yes, I am aware that most of these owners are not ready to cooperate (narcissistic narcissists) and are completely selfish, but usually 2-5% can help you in your choice. It is for the sake of finding adequate owners who cooperate and help that it is worth writing on the forums. You help others by responding to the questions posed, someone may answer you by checking the things you need.

lmao

Quote from: NikoB on March 09, 2024, 11:29:49You will not find in the text of this "review" a single mention of PL1/PL2 in different modes and under complex loads.
because it's useless

Quick Reply

Name:
Email:
Verification:
Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview