Question Lenovo IdeaPad S540 testing - 4800U vs 10710U+MX250

lightmanek

Senior member
Feb 19, 2017
387
754
136
What strikes me is that you can set AMD model to one or two tier lower TDP and still get roughly the same or better performance in a wide spectrum of applications! This means that if you would try to optimize results for equal performance, all of the sudden 4800U could last almost twice as long on batteries giving very similar user experience when mobile!
 

gorobei

Diamond Member
Jan 7, 2007
3,654
980
136
i did the support bot chat on the lenovo site, and the ideapad c-series (with pen support) will be available in july or august. will probably be my next purchase.

i was hoping for a 2in1 with the cooling vent in the hinge but the lenovo checks off all my needs aside from the soldered on board memory.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
Of course because the Intel laptop has a GeForce MX250, it wins the gaming benchmarks. Typical.

The dGPU disadvantages the Intel system in lighter battery life tests. The switchable graphics aren't perfect and the dGPU still uses a bit of power. At least the tested applications are working fine with it.

Renoir seems quite a bit better on CPU though.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
Of course because the Intel laptop has a GeForce MX250, it wins the gaming benchmarks. Typical.

MX250 win only in Sleeping Dogs,
In Civilization at 15W TDP, 4800U and MX250 are equal with the 4800U having higher minimum fps
In Dawn Of War at 15W TDP, 4800U is 7% slower at the avg fps but 19% faster on the minimum fps

I dont consider this as MX250 wins in gaming benchmarks.
I do like to see more games benchmarked though as only three games is not enough to draw final conclusions.
 

LightningZ71

Golden Member
Mar 10, 2017
1,627
1,898
136
In general, on mobile, without heavy TDP (power/thermal) constraints, it looks like the 4800u should be roughly on par with the MX250 in most situations, maybe just a bit behind. I would expect that the 4700U should be able to keep up with the MX230. The 4600u/4500u seem to have taken a bit of a GPU performance regression from the 3500U in a few of the benchmarks that I've seen, though, again, its very limited.

Essentially, if you find a 4500/4600 with a dgpu that's less expensive than the 4700/4800u and have any intentions of doing low spec gaming, the 4500/4600 is the better buy. If you find the 4700/4800 cheaper, then unless the 4500/4600 has a substantially better dGPU, the 4700 is the better buy. I suspect that, for purely mobile gaming though, where you have to run just off the battery, the 4700/4800u will give you the best of all worlds if you balance performance, battery life, portability and cost. I think that, if people are willing to game at a rendering resolution of 720p and use RIS with upscaling to 1080p, the mobile experience should be just fine for most people.

I do wonder how good the 4800u can be? We haven't gotten a good round of benchmarks off of a system that's outfitted with "dual" channel LPDDR4X 4266 in a system with proper TDP up and cooling. I suspect that it'll get a good 5-10% higher results in that configuration...
 
  • Like
Reactions: Tlh97

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
MX250 win only in Sleeping Dogs,

It wins in everything @ 25W. Gaming performance for the 4800u doesn't scale up performance-wise that much with an expanded power budget. Introducing a dGPU lets the Intel system stretch its legs. Both the OEM and Intel are aware of this fact. Benchmarks like that keep Intel in the game.

If the OEM had even a niche, low-volume version of that laptop with an MX250 and 4800u in it that they could include in their review, I'm confident the 4800u would win every one of those game benches. It is very important for Intel to prevent that.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
It wins in everything @ 25W. Gaming performance for the 4800u doesn't scale up performance-wise that much with an expanded power budget. Introducing a dGPU lets the Intel system stretch its legs. Both the OEM and Intel are aware of this fact. Benchmarks like that keep Intel in the game.

If the OEM had even a niche, low-volume version of that laptop with an MX250 and 4800u in it that they could include in their review, I'm confident the 4800u would win every one of those game benches. It is very important for Intel to prevent that.

4800U doesnt scale up at 25W TDP because this Laptop uses DDR-4 3200 (51.2GB/s) and not LPDDR-4X 4266 (68.3GB/s)
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
Okay. And? You know as well as I do that all Lenovo had to do was put an MX250 on that laptop with the 4800u, and the 4800u becomes fastest in everything.

Not the same, in one place you just add LPDDR4X, on the other hand you have to add the dGPU.

BOM is way higher with the dGPU and performance will be almost the same between the two versions, the APU + LPDDR4X model will be cheaper to manufacture (cheaper selling price or higher profits)
The laptop with the LPDDR4X will have even lower idle power than the one with DDR-4 3200 thus increased idle battery life.
Also, the laptop with the APU + LPDDR4X will have lower thermals than the one with the dGPU + DDR4
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Not the same, in one place you just add LPDDR4X, on the other hand you have to add the dGPU.

BOM is way higher with the dGPU

You are overthinking this problem.

Intel knows their CPU is inferior (10710U vs 4800u). If their CPU is paired with twice as much RAM, a dGPU, and a faster NVMe drive, they win some benches (gaming and WinRAR) and drive sales. BoM is not important. They must drive the perception that their CPUs are still premium products that are more-desirable than "cheap" AMD solutions. AMD is stuck with half as much RAM and a slower NVMe drive (which ironically gets twice the capacity but oh well).

This is all about marketing and perception. If you start talking about DDR4 vs. LPDDR4X and so forth then you've already missed the point.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
You are overthinking this problem.

Intel knows their CPU is inferior (10710U vs 4800u). If their CPU is paired with twice as much RAM, a dGPU, and a faster NVMe drive, they win some benches (gaming and WinRAR) and drive sales. BoM is not important. They must drive the perception that their CPUs are still premium products that are more-desirable than "cheap" AMD solutions. AMD is stuck with half as much RAM and a slower NVMe drive (which ironically gets twice the capacity but oh well).

This is all about marketing and perception. If you start talking about DDR4 vs. LPDDR4X and so forth then you've already missed the point.

I completely disagree,

LENOVO Yoga Slim 7 ultra slim with Ryzen 4800U and LPDDR4X

Leaked gaming benchmarks suggest its faster than MX250 and very close to MX350

edit: MX350 is a 25W TDP dGPU

 
Last edited:

uzzi38

Platinum Member
Oct 16, 2019
2,565
5,573
146
4800U doesnt scale up at 25W TDP because this Laptop uses DDR-4 3200 (51.2GB/s) and not LPDDR-4X 4266 (68.3GB/s)
If only...

1:1 mode using LPDDR4X is impossible. Renoir still runs into the same ~1900mhz cap on 1:1 FCLK, and at stock it has the same limitations as Matisse for RNR-H. >3600mhz DDR4, assuming you can force memory to run at that via the BIOS. Meanwhile for -U that limitation is at >2666mhz DDR4 instead. In the case of a -U laptop like the point in question here, both bandwidth's are halved the rate mentionned above due to running FCLK in 2:1 mode.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
If only...

1:1 mode using LPDDR4X is impossible. Renoir still runs into the same ~1900mhz cap on 1:1 FCLK, and at stock it has the same limitations as Matisse for RNR-H. >3600mhz DDR4, assuming you can force memory to run at that via the BIOS. Meanwhile for -U that limitation is at >2666mhz DDR4 instead. In the case of a -U laptop like the point in question here, both bandwidth's are halved the rate mentionned above due to running FCLK in 2:1 mode.

source for all that ??
 

Gideon

Golden Member
Nov 27, 2007
1,608
3,572
136
I don't get why the bandwidth should be halved? (For Infinity Fabric yes, but not for memory). But the fact that Renoir is still limited to a FCLOCK of 1800-1900 is definitely true. The chinese review that tested memory in detail confirmed it (maybe someone cares to look it up?)
 

lobz

Platinum Member
Feb 10, 2017
2,057
2,856
136
It wins in everything @ 25W. Gaming performance for the 4800u doesn't scale up performance-wise that much with an expanded power budget. Introducing a dGPU lets the Intel system stretch its legs. Both the OEM and Intel are aware of this fact. Benchmarks like that keep Intel in the game.

If the OEM had even a niche, low-volume version of that laptop with an MX250 and 4800u in it that they could include in their review, I'm confident the 4800u would win every one of those game benches. It is very important for Intel to prevent that.
it means when the dGPU _alone_ uses 25W
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
I don't get why the bandwidth should be halved? (For Infinity Fabric yes, but not for memory). But the fact that Renoir is still limited to a FCLOCK of 1800-1900 is definitely true. The chinese review that tested memory in detail confirmed it (maybe someone cares to look it up?)

What Chinese review ??
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
I completely disagree,

Why? You're ignoring everything I'm saying rather than telling me anything that makes sense. The linked review in the OP shows the MX250 winning. You can't get around that with some leaked info that "suggests" things. We have a straight-up comparison of MX250 @ 25W beating the 4800u's iGPU no matter that the power configuration. Never mind that you are missing the point.

You are not the review site. NextLabs501 is the review site. They already produced the benchmark wins for Intel, which is exactly what Intel needed for their chunk of silicon not to be rendered completely irrelevant. Perception. That is the name of the game.

it means when the dGPU _alone_ uses 25W

Intel takes any win they can any way they can. You know this puts the Intel rig at a disadvantage wrt battery life, and so do I, but the pretty bar graphs don't go into that until later in the review.
 

IntelUser2000

Elite Member
Oct 14, 2003
8,686
3,785
136
it means when the dGPU _alone_ uses 25W

This is not exactly the case. CPU+GPU TDP is not CPU+GPU in systems that use the dGPU. Without needing to run the iGPU, the TDP of the CPU decreases, especially on systems that use the low end dGPUs where the system is bottlenecked by the GPU and CPU is not running at full.

The actual power will be higher than the iGPU system but it'll be 25W + some fraction of CPU TDP rather than 25W + 15W.

Also battery life is crap in laptops when gaming anyways so this argument is mostly moot.

This is how Nvidia managed to win a trifecta of power/price/performance over Intel's Iris Pro laptops back in 2013.
 
  • Like
Reactions: Tlh97 and lobz

Gideon

Golden Member
Nov 27, 2007
1,608
3,572
136
What Chinese review ??
Sigh. Just One time i didn't provide a link because it was 2AM here and it was a obscure hard-to-find early review and instantly I must be pulling info out of my ass right?

Here, referenced by no other than @chiakokhua:
There's a very detailed analysis of mem-speed and FCLK there, which is quite readable with Google translate (lots of AIDA64 pictures at source too):

The 4800H uses 3200 memory, so the core display momery clock is 1600MHz; and the 4800U I use LPDDR4X-4266, the core display momery clock should be 2133MHz That's right, why is it only 1200MHz when running, and even falls to 1066MHz?
Thinking for a long time with doubtful thoughts, suddenly an expert woke me up-this may be caused by the IF bus frequency division of the ZEN2 architecture
...
Because the IF bus frequency is limited to about 1800MHz, the higher memory frequency causes the IF bus to be divided, thereby ensuring the stability of the CPU. The notebook platform is divided at 3600MHz, while the desktop is 3733MHz, slightly difference
Before and after frequency division, the most significant effect is the memory response time. At 3733 / 4266MHz, the delay is more than 120ns, while at 3600MHz, it is only 105ns. The difference is very significant. Of course, LPDDR4X still can't do the delay of DDR4, but the overall effect it is good


Besides it should be expected that zen2 runs like zen2 not some magical fairy dust just because it's on a single Die