Asus ROG Strix GL702ZC / G702ZC / S7ZC with Ryzen 7 1700 8-core CPU and a Radeon RX580 GPU

Only about 10

 

And I wonder how many more years would 4c be a standard if AMD didn't come along and kick Intel in the behind (and accidentally shattered it in the process).

 

Probably another 2 at least. Till Cannon Lake.

 

Maybe.
At any rate, Vega in mobile is probably more than doable.
With (admittedly) unconfirmed benchmarks of Vega 56 getting within a hair of 1080 with 210W TDP (reference model at high stock voltage to increase yields), I'd imagine we can easily see a mobile version at 150W or less.

Asus was able to reduce RX 580 TDP from some 180W to 65W. Granted, that's with core clocks at 1077Mhz and boosting beyond 1200 MhZ stable (so, close to stock desktop 480).
If they can repeat the process with Vega (and there's really no reason they can't since all AMD gpu's carry a degree of suitability for undervolting), plus binning and better silicon... yeah, I'd imagine that reducing the TDP to say 120-130W while leaving everything else as is should be possible.
But in reality, we need to see how Vega performs, and just what all of its features can do (including power saving ones).

 

AMD takes awhile to put out a good card, but when they do... it's worth it. Something tells me Vega will be as well. It's usually on par with Nvidia's current generation at a slightly lower cost. Nvidia usually trumps out later, but they are worthy cards nonetheless. And if they can be paired with Ryzen...

Let's be honest, the 580 is hardly any more powerful than the r9 290x released 4 years ago.

For some reason I thought the Vega 56 wasn't the flagship, rather more of a midrange model. Is that incorrect information?

 

You are right, Vega 56 is the lowest end Vega (rumoured to be 20% ahead of the 1070) whereas Vega 64 is the high-end model available with multiple configs.

As for the 580 - it's basically a 290X for less than half the original price. It's also more efficient and has more VRAM. Basically, it's making gaming cheaper, not making overkill hardware (which imo is what 90% of users are looking for - to spend less money on what they will actually use rather than buy a 1080 Ti that can only be taken advantage of in a few situations)

 

Condisdering the Asus Stryx laptops with all AMD hardware, I would imagine that RX 580 will last us a good while longer.
It's more than capable of playing games maxed out at 1080p and do similarly with 2K with playable framerates.
For 2K (which will require outputting to an external 2k TV or monitor), if we drop the in-game quality from 'ultra' to 'high', we really won't lose any visual quality (or lose an unnoticeable minimum) while increasing playable framerates by a great deal.

Plus, if the RX 580 in those laptops is modular, we might be able to upgrade to Vega or Navi if they are released in suitable formats.
I would imagine that Navi will be using full blown Infinity fabric to interconnect several Vega dies - unless we are also talking about architectural changes (AMD should really try seriously improve perf. per watt in a similar way that Nvidia did with Maxwell).

 

Well, to be honest, efficiency is already superior to Maxwell (a RX 580M performs like a 970 (145W) while using less than 80W) - it's mostly because of the high clockspeeds that the power draw on Vega is as it is. GloFo's process really doesn't play nice with high clockspeed.

 

There's that, and AMD's tendency to overvolt their reference GPU's so they can increase yields (producing higher power draw and potential thermal issues that prevent the GPU from boosting properly - hence why undervolting fixes these issues, usually increasing performance by about 5% and reducing overall power draw).

The non-reference 480 is already more efficient when properly undervolted vs 1060.

We know that GPU's from partners can come with lower voltages than reference GPU's and better silicon (usually)... but the trade-off is that the partners then increase clock speeds for a minute gain while wrecking efficiency in the process.
AMD needs to dial back on the clocks for this particular process after a 'comfortable ceiling' to still maintain good efficiency ratio... their OEM partners should do similar, or ensure that they modify voltages to increase efficiency.

Heck, even Vega FE was undervolted to 1.12V stable across all loads and it managed to reduce power draw by up to 90W - which is no small feat, and it allowed the GPU to boost and maintain 1600MhZ for as long as it needed it (this is on air of course).
Plus, that is an early silicon GPU if I'm not mistaken - and it has a lot of computational hardware active which makes it power hungry for pro loads - which is probably why the TDP's in relation to gaming GPu's are not different, because they have same hardware, but running off different software.

I've been reading some Reddit comments from Polaris users, and they were able to hit stock clock rates on RX 580 (above 1300MhZ) on very low voltages (we're talking about 1.1V or less).

We still don't know how Vega will ultimately perform and what all its features when active can do... if they do anything like that in the first place, and of course, we know even less of its voltages when coming from other partners.

 

I want this.

 

Sooo, buy it?

 

I broke T-T

Im just hoping its around long enough for me to pick up if and when available in the US

the other thing that concerns me is the 4GB VRAM but otherwise I really like what I see. Though I already know I would be cutting out 2 holes for air intake. Im not sure why Asus keeps on making them like that...but ah well no matter can be easily remedied.

 

First GL702ZC with Ryzen 7 1700 review:
www.ultrabookreview.com/17680-asus-rog-gl702zc-reviews/

 

Fire Strike scores seem a little low. The Omen 15 gets about 11500-11700

 

Too bad Asus couldn't put a thermal system like this on the 702 Intel/Pascal variant. Had they used something like this, they would probably stay below 85.

 

The games they picked are games that favor Nvidia and are worse performers for AMD.

There are many games on either side of performance advantage available, but I suppose some reviewers game libraries available for testing will favor one brand or the other depending on their personal daily use.

I'd check out the desktop 580 / 1060 comparisons and find the games with best AMD performance and then run those in even number with the ones that favor Nvidia for a fair comparison.

And, I did, so here is a Hardware Unboxed review that revisits 1060 vs AMD after new drivers and a new rev of the RX 480 to the RX 580. Showing the original 12% advantage of the 1060 dropped to 1% later with the RX 480, and now (5/17) showing the RX 580 as 2% faster overall.

The review, Can the RX 580 Dethrone The GTX 1060? Yep, We're Doing It Again!, has a nice chart showing the mix of games faster on the RX 580 vs faster on the 1060:

So remember, the games chosen for benchmarking Nvidia vs AMD can skew results one way or the other depending on those choices, either way the RX 580 + Ryzen mobile is looking great, maybe even super great

 

55dB (at head level under load) seems too loud to be successful, that's about as loud as poorly reviewed desktop graphics cards, which can be pushed farther away from the user. A 45 C sweaty keyboard is also uncomfortable.

A problem seems to be that the GL702ZC design tries to dissapate much more heat with the same size radiators as the already overwarm GL702VT and GL702VM did (125W -> 215?), so the fans have to run much faster and louder. I thought Asus would be able to make the radiators deeper (front-to-back), wider, and/or taller. But depth and width appear the same, and the 34-28=6mm height increase accommodates the additional layer of heat pipes at the radiator, so the radiator is about the same size. [Small Ultrabookreview article typo: height 34mm is about 1.34in (not 0.67in).]

Another issue is that the fans are not positioned over vents in the bottom. Apparently they are designed to draw air from over (both sides of?) the circuit board, which helps cool other components and the heat pipes, but preheats the air so the air is not as effective at cooling the radiators. If it draws air from the rear center grill, it could be recycling some exhausted heat, especially if the back is near a wall. I expect some owners will try adding additional holes to improve the ventilation, as they did with the Lenovo Y700, even if the GL702ZC isn't throttling.

(I wonder if the heat pipes that extend past GPU to other components might stop working when the GPU is hot. When the GPU is hot, it might evaporate all the liquid near it, so insufficient liquid might get past the GPU to the additional components, so they won't be cooled by evaporation.)
[As I understand, a heat pipe contains a small amount of liquid with low boiling point that vaporizes with heat. At a hot component like the GPU, the liquid turns into a gas and the phase change absorbs energy. The gas disperses through the heat pipe. The gas near the cool radiator condenses back to a liquid and the phase change releases energy. (The released energy heats the radiator fins which are cooled by air blown by the fan.) The liquid then makes its way back toward hot components through capillary action (wicking).]

P.S.

Technically, this article is still a preview of a preproduction model
(as was the Ryzen 1600 + RX 580 preview last month: Asus ROG Strix GL702ZC benchmarked).

Unhappily, it looks to me like the preproduction preview hardware hasn't changed in the 6 weeks since the previous preview, so the hardware design might be frozen.

But Prime95 temperatures did improve from 95C 3.3GHz (Ryzen 1600 GL702ZC in July) to 75C 3.2GHz (Ryzen 1700 GL702ZC in August).

 

That's the same sound level measured by @Darkhan for his new Max-Q 1080 Zephyrus at load.

He had 55dBA measured at head level here in this post, but it looks like he edited that detail out...

Darn, it looks like he edited out all those 50-55dba entries in his posts...

What's up with that @Darkhan ? If it's true, and you posted it, why add to the Max-Q deceptions by editing out the useful details?

I was quite impressed that you posted the honest real measurements even though it disproved Asus' Max-Q's claim of 39dBA - what happened?

 

Asus does this frequently, keeps the demo mules running around the review sites and trade shows until the production units are ready to ship, and then user reports often come first. It happened with the water-cooled Asus laptops too.

I'm sure Asus has a number of test beds being used on site for optimizations, and often those aren't pretty, only functional - maybe not even in a laptop case.

It is exciting to think of the Ryzen + RX56 and Ryzen + RX580 models in production

 

The actual TDP of the components in that is 65W (CPU) + 80W (GPU) for a total of 145W. The 580 is downclocked and we know the Polaris/Vega arches are overvolted due to the stock OCs thus using far more power than they would if you dial back the OC and voltage.

As for the heat and noise - I mean it's still a full 8-core CPU that performs very closely to Intel's 7900X (which is around 200W under load IIRC) so yeah, it does need better cooling than the silly HQ quad-cores that can't even hold max turbo due to TDP limits.

 

In Germany, Computerbase ( de) and NotebookCheck ( de) expect Asus GL702ZC preorders to begin,
from 8 models (3 binary choices), with delivery to begin about 21 September.

DE	UK	US		AM4 Socket	DDR4 Slots	(soldered GPU)	(soldered)	M.2 Slot	2.5" Slot
			+0
€ 1400				Ryzen 5 1600	8 GB	Radeon RX 580	4 GB	(empty)	1 TB SSHD
			+1
€ 1500				Ryzen 5 1600	16 GB	Radeon RX 580	4 GB	(empty)	1 TB SSHD
€ 1500				Ryzen 7 1700	8 GB	Radeon RX 580	4 GB	(empty)	1 TB SSHD
			+2
€ 1600	£ 1400			Ryzen 5 1600	8 GB	Radeon RX 580	4 GB	256 GB SSD	1 TB HDD
€ 1600				Ryzen 7 1700	16 GB	Radeon RX 580	4 GB	(empty)	1 TB SSHD
			+3
€ 1700				Ryzen 5 1600	16 GB	Radeon RX 580	4 GB	256 GB SSD	1 TB HDD
€ 1700				Ryzen 7 1700	8 GB	Radeon RX 580	4 GB	256 GB SSD	1 TB HDD
			+4
€ 1800	£ 1590	$ 1500 +tax		Ryzen 7 1700	16 GB	Radeon RX 580	4 GB	256 GB SSD	1 TB HDD

(UK preorder prices are from Overclockers. US preorder price is from Excalibur)
(A choice of screen that Bit-tech predicted for the UK market is not an option for GL702ZC preorders.)
(Adding AMD models occupies more e-catalog 'shelf-space', fighting for popularity?)

 

So about 1700 euros (1800$) for a fully loaded R5 1600 model? I mean, it's pricey but it's 50% more CPU performance than any other laptop on the market and the GPU is quite capable as well. Though this is + VAT so I imagine the US price will likely be about 1400$ which is much more reasonable.

 

50% more Cpu power or not, fully locked down, and cheaper Cpu than the locked down 7700hq isn't a good deal. And Soldered AMD gpu means a big no!!

 

Cheaper or not - the 7700HQ can't even compete. It performs worse than a R5 1500X. As for the GPU - It doesn't matter who makes it as long as it performs well.

 

The Radeon 'RX 580' GPU in this GL702ZC prototype image appears to have only 7 of the normal 8 memory chips.

If so, this has consequences and raises some questions:

The reduced core clock rate (1077MHz vs. 1257MHz) results in potentially less memory bandwidth consumed. Lower memory bandwidth consumed has been translated to cost saving via one less memory channel and GDDR5 chip. One less channel means a 7 × 32 = 224-bit memory interface rather than an 8 × 32 = 256-bit interface. Bit-tech reported the memory clock is not reduced, so the peak memory bandwidth may similar to an RX570. (RX570 peak bandwidth: 224 GB/s. RX580 peak bandwidth: 256 GB/s. 224/256 = 7/8.) But RX570 does it using a lower memory clock rate (1750 MHz vs. 2000 MHz), not fewer channels.

With fewer channels, this GPU cannot be expected to perform like a normal RX580 on memory-bound computations, despite the identical memory clock. Interleaving and striding will behave differently.
​

Does the production laptop use the same approach?

How is the memory divided among the 7 chips?
a. 7GB = 7 × 1GB
b. 3.5GB = 7 × 512MB
c. 4GB = (6 × 512MB) + (1 × 1GB)

How is memory interleaved among 7 channels rather than 8?

​

Shall we call this configuration an 'RX580M' to differentiate it from a full 'RX580'?

Do other 'RX580' laptops also use this 'RX580M' design to reduce memory chips, or is this an Asus cost- and space-saving measure?

If the memory had all 8 chips and was downclocked instead, would that noticeably reduce the fan noise?

 

GDDR5 is not really that power hungry and I doubt 1 chip more or less would impact cooling or fan noise at all. Especially at the clockspeeds given.

 

Asus ROG Strix S7ZC (GL702ZC)

Two more preview reports, from visits to AMD's booth at ChinaJoy 2017, Shanghai, end of July.

[For China, apparently Asus shortens the series from 'GL702' to 'S7', so
"ROG Strix GL702ZC" becomes "ROG Strix S7ZC".]

1. [ Cotton Zone] blog, 2017-07-30.
https://zhuanlan.zhihu.com/p/28298470 (viewed in Google Chrome with translation)

The report says the prototype screen was a 60Hz Freesync display.
This report refers to the GPU as an "mRX580".

Cinebench R15 single and multithreaded scores matched desktop Ryzen 7 1700.
144 single threaded matched 144 for i5 7300HQ.
1407 multithreaded was near double 742 for i7 7700HQ.
[Apparently Cinebench threads benefit from cache sharing.]

On TombRaider with extreme high effects, 1080p, DX12:
57 Avg FPS was close to a 59 for GTX1060.
28 Min FPS was off a 44 for GTX1060, maybe due to 4GB memory limit.

AIDA64 stress test showed no CPU throttling while holding 3.2GHz on all 8 cores, though the author wondered about the 57.5 degrees for 77W. [No word on how long it ran.]

2. ZOL Home, dated 2017-08-05
http://nb.zol.com.cn/649/6490253_all.html (viewed in Google Chrome with translation)

Cinebench R15 single and multithreaded scores
120 single threaded
1337 multithreaded

Fritz Chess Benchmark
38.16 ratio [compared to Pentium3 1.0GHz]
18318 rate [kilonodes per second]


[For comparison, previous Cinebench R15 multithreaded scores:
1129 reported by bit-tech reported on July 13.
1407 ultrabookreview also reported on Aug 23.]

 

i am just happy to see AMD back in the ring !! that being sead this is my first intel i7 7700 and its dam snappy

 

Just as a heads up, it's up for pre-order on excaliberpc

http://excaliberpc.com/661022/asus-rog-strix-gl702zc-wb74-17.3.html

For what it is, it's not bad, but I'm going to wait for a 120hz version , and see how well a good repaste affects temp values before I decide if I wanna trade in my 502vs.

 

Notebookcheck.com ( de) has started testing an Asus ROG Strix GL702ZC. No text yet, but some components listed and some benchmark graphs are posted. (Too bad they don't have the RX580 Omen 17 to compare mobile GPU cooling and performance with 7 memory chips.)

 

How'd they achieve that price? Partly with
Single channel memory default for an 8 core CPU.
SATA III SSD (in a slot capable of NVME x4).
Click "customize" to

• increase memory,
  
• change storage (many options, The largest storage options more than double the price, perhaps targeted for video editors or big data bases.),
  
• upgrade OS to Pro, or
  
• select ROG accessories.

Clicking "preorder" produces a page where the free ROG-branded accessories are included with zero cost. Clicking "customize" produces a page that advertises some "includes free item" ROG-branded accessories, but they are not listed on the manifest, and adding them increases price.
(So maybe - they are free only in the bundle configuration? or - "free" is short for "free shipping", not free for preorder? or - Excalibur's web site is misconfigured?)

Also the customize page says it ships in 3-5 days, but that is probably incorrect if they are only accepting preorders at this time.

The RMA page says low prices are partly due to no refunds, returns forfeit $25 + 15% (so ~$250), return sender pays return shipping cost.

 

This is no different than any other laptop when boutique sellers decommission parts to lower the price, it's a common technique, and much appreciated for those trying to afford a laptop and upgrade parts later

 

Fact of the matter is, even more loaded notebooks still have options to upgrade when it comes to boutiques. I for one like it since it's cheaper to diy (although i'll probably let an expert do the repaste if i get one).

 

Review is live, still not complete, especially thermals, but there is more info now, such as benchmarks and translated to english.
https://www.notebookcheck.net/Live-...zen-7-1700-Radeon-RX-580-Laptop.247548.0.html

 

Intel might introduce desktop cpu on laptops after seeing Ryzen threat.

 

Some OEMs already do that but those laptops are absurdly massive and require a delid otherwise they risk melting themselves.

 

Laptop Upgradable LGA/PGA CPUs http://forum.notebookreview.com/threads/cpu-upgradeable-laptops.805499/#post-10538481

 

Same AMD Ryzen 7 1700 configuration with different cooling solutions:




https://www.notebookcheck.com/Live-...zen-7-1700-Radeon-RX-580-Laptop.247470.0.html
http://www.ultrabookreview.com/17680-asus-rog-gl702zc-reviews/

 

With this fan noise (according to NBC), it doesn't have any chance with me.

 

Might be just me but I can't see any fan noise data. Can you point me to the correct graphs? thanks.

 

I looked to the German review: https://www.notebookcheck.com/Live-...zen-7-1700-Radeon-RX-580-Laptop.247470.0.html

Still no comments but I think figures shows quite a lot.

 

I guess heavy load does make it a loud machine. Though it also has 42dB(A) listed as minimum load noise so I guess it depends on the load. I imagine it will be around 50dB(A) in most cases. Still loud but along the lines of most others.

 

Doesn't look promising. Looks at this GPU Temps.

Sent from my SM-G955U using Tapatalk

 

Maybe the R5 1600 + 580 model will fair better but yes, it's a hot machine.
I wonder how a repaste and undervolt will change things, though.

 

I don't know but do the engineers at Asus not know how to design a proper thermal management system?

Sent from my SM-G955U using Tapatalk

 

Which noise numbers bother you? They look average to low to me. Here's the GT73 review, compared to those noise measurements they look similar, which is what I would expect:
https://www.notebookcheck.net/MSI-GT73VR-7RF-Titan-Pro-Notebook-Review.189710.0.html

 

Yeah, It's scary that engineers working in team can't do it any better. One engineer can approve it and mean it's good enough, but more than one engineer working on projects like laptop design.

 

The GPU temps are 81c/85c on a heavy load test, not bad, certainly not worth those comments.

Coming from the Clevo world I'd expect you all to be very happy with those numbers

 

Personally I've never owned a Clevo. I know it's not good to base your purchases off of looks but damn they are ugly. A little to ugly for me. That being said, the PA70 is the first Clevo design I've liked.

I've always been an Asus/MSI type of guy but lately, Asus is worrying me with these thermal designs. I was all set on a GE73vr but I don't game as much as I used to and I feel like the 1070 would just be going to waste considering the only games I play are Counter Strike, Hearthstone and Overwatch. I think the added thermals and fan noise of the 1070 would be pointless. The 1060 version is too expensive at this point.

Honestly I don't know what I'm doing at this point. I've been researching my next purchase for the past month and I'm constantly going back and forth between multiple models. I've came to the conclusion that there are no perfect laptops but I don't know what I'm waiting on.

Sent from my SM-G955U using Tapatalk