News:

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

Main Menu

Post reply

The message has the following error or errors that must be corrected before continuing:
Warning: this topic has not been posted in for at least 120 days.
Unless you're sure you want to reply, please consider starting a new topic.
Other options
Verification:
Please leave this box empty:

Shortcuts: ALT+S post or ALT+P preview

Topic summary

Posted by william blake
 - January 28, 2020, 13:29:00
Quote from: _MT_ on January 28, 2020, 09:35:13
That's why I appreciate that notebookcheck performs throttling and sustained load tests. And you can put a benchmark right after such a load and check the difference (when the laptop is well baked vs. cold). Which leads me to another thing and that is benchmarking laptops primarily connected to a wall. That just doesn't make sense to me. It makes sense for big beasts which are not meant to be used on the go (they're transportable, not mobile). There is no way their battery would be able to keep up anyway. It's a machine you can bring with you to a weekend house instead of packing a desktop, not something you unpack at an airport.
most of hardware testing places over the internet are not very good. most has pretty weak methodology. my russian ixbt.com is the best overall. they created their own methodology and even benchmarks with the cooperation with users for dozens of hardware and all other home electronic stuff.
i like notebookcheck for huge database and retro non-phone design :)

Quote from: _MT_ on January 28, 2020, 09:35:13
But that doesn't explain his problems with his desktop.
desktops is a different animal. mostly because you are a creator, its easy to avoid some annoying bullshit from manufacturer. also tdp. for example, fast igpu make sense.
1.enough tdp room for both cpu and gpu.
2.size matters. and desktop discretes are big.
Posted by _MT_
 - January 28, 2020, 09:35:13
Quote from: william blake on January 28, 2020, 00:23:10
intel is a monopoly for a looong period. so it is exactly intel's fault.
Why should a chip maker be responsible for the finished product? Yes, they could motivate such a development. Just as they motivated the development of ultrabooks or what they did with Centrino before that. But they typically only do that with the view to increase sales. As long as we keep buying it, Intel doesn't have to care about the incompetence of manufacturers. And is it really incompetence or conviction that this is what the market wants (what they can successfully push out)? Frankly, there is not a single laptop on the market that ticks all the boxes for me (in the relevant category; obviously, ultrabook won't fill my workstation needs and vice versa). But I'm used to being the minority. I know what I want and I understand the compromises, unlike a typical consumer. Some are close, but all have annoying quirks. The fact that I hate 16:9 aspect ratio and glossy screens doesn't help. :-) What really annoys me is that even workstations can look like the manufacturer doesn't care. It's almost like "it's thick, we don't have to try hard." I don't want to be looking at gaming laptops.

That's why I appreciate that notebookcheck performs throttling and sustained load tests. And you can put a benchmark right after such a load and check the difference (when the laptop is well baked vs. cold). Which leads me to another thing and that is benchmarking laptops primarily connected to a wall. That just doesn't make sense to me. It makes sense for big beasts which are not meant to be used on the go (they're transportable, not mobile). There is no way their battery would be able to keep up anyway. It's a machine you can bring with you to a weekend house instead of packing a desktop, not something you unpack at an airport.

But that doesn't explain his problems with his desktop. I mean, it's not like he is running a W-3175X. That is a truly ridiculous processor. You can certainly ask "what were they thinking?" With a 65 W TDP chip? If you can't cool that, the joke is on you. If I had to guess, my first guess would be a case with poor airflow. Once it heats up, you're screwed. I don't use box coolers, but surely they're not that bad, are they?
Posted by william blake
 - January 28, 2020, 00:23:10
Quote from: _MT_ on January 27, 2020, 20:04:33
That's not really Intel's fault.
intel is a monopoly for a looong period. so it is exactly intel's fault.
imagine intel initiative "extra heat pipe and bigger quieter fans". right now 40w and 40 db is a common thing. because oems dont care at all and people are pathetic slim bezel consumers. not a slim lower bezel tho, 5cm lower is fine. there is no excuses if a laptop with 15w is noisy and throttling all the time. one heat pipe and 20y.o. 5k rpm fan. just imagine 2 heat pipes and 2k rpm bigger fan. no noise no throttling. yes your laptop would be 1mm thicker. maybe. what a disaster.
anyway, the question is how reliable and informative all these benchmarks are? because, you know, sometimes 50w sometimes 10w but we are talking about 50w performance only.
Posted by _MT_
 - January 27, 2020, 20:04:33
Quote from: Rico Mico on January 27, 2020, 18:19:58
With this I mean that, with Intel you think you have a good CPU when  the reality is something else. They play with bursts and intelligent detection of single/ multi core and reduce speed so that when working the CPU burst for a little while on heavenly speeds until the fall to the real lousy speeds. Also with spectre & Co correction,  I lost around 8-12% of speed.

That's not really Intel's fault. Yes, laptops are primarily heat limited. And secondarily battery limited. If you have something like a 50 Wh battery and you want it to last a whole work day, your budget is about 6 W on average. That's laughable. The extra power limit is a bit like a turbocharger in a typical, non-performance car. It's there when you need it. For a typical user, doing something like web browsing, those bursts actually help. It's not a constant load. If you want to hammer at something, you need something heavier duty. First cooling, second power and only then comes the chip. If you can't cool it or feed it, it's pointless. Under these constraints, sustained performance is about efficiency.

Same principles apply to a desktop. They're just usually less of a problem. My desktop can keep full boost indefinitely. Because it has the cooling to match. And yes, when your workstation draws 600 W from the wall, it's a neat little space heater. :-) I like what AMD is doing, but going AMD won't solve crappy cooling. That's on whoever made the computer.
Posted by Rico Mico
 - January 27, 2020, 18:19:58
I have an Asus with 8 th gen chips, 15 W TDP, in reality it bursts up to 44W on short periods, 25W sometimes and then - when the laptop gets really hot- less than 10W (though the temperature is still 65-70°C) and perhaps the average is...15W. The thing is, when I play or render something at first you get an ok speed, after 10 minutes... disgrace.

On my desktop an Intel CPU (65W). Short periods are good and fast. Continuous medium work and the fan starts to speed a lot and speed is not stable anymore; With heavy work the desktop gets very hot and forget bursts (and good speed).

With this I mean that, with Intel you think you have a good CPU when  the reality is something else. They play with bursts and intelligent detection of single/ multi core and reduce speed so that when working the CPU burst for a little while on heavenly speeds until the fall to the real lousy speeds. Also with spectre & Co correction,  I lost around 8-12% of speed.

So my next PC will get am AMD CPU/APU and perhaps GPU too. I am tired of the Intel/ nvidia combo, always pressuring manufacturers to use their hardware and let the slow SSD, single channel stuff for AMD.
Posted by william blake
 - January 27, 2020, 14:54:33
Quote from: Vaidyanathan on January 27, 2020, 07:39:57
I think on a CPU level, things are leveling out and AMD has to be commended here. But I am still wary about what kind of laptops OEMs actually push out. Intel's hegemony can't be discounted all of a sudden. We did see some cool designs at CES and about 100 designs are on the anvil, but for a common person, it is still Intel Inside, at least when it comes to laptops or ultrabooks.
totally understandable view from a "common persons". if 9 of 10 laptops has intel inside, then laptops are intel inside. nevertheless, laptops are much easier than servers and having a superior products helps a lot with fast growing. as i said previously, amd's 2019 and 2020 mobile lineups are night and day. i expect big changes in mentality of a common person this year :) servers is a sad story, not laptops. right now, corporations are buying pricy, hot and slow intel here and there. and we are talking about really big differencies in product quality. not just slightly but times better. i need some good readings about this situation, i have no idea how it happens. times better but i will buy the worst one-impossible in laptops
Posted by Vaidyanathan
 - January 27, 2020, 07:39:57
Quote from: william blake on January 27, 2020, 04:51:45
Quote from: k on January 27, 2020, 03:52:20
firstly, with better brand name and deeper distribution chain, intel is always with premium configuration.
kinda common knowledge. but not about technology or products.
Quotesecondly, most light duty software are not mutlithread ready and heavy one's users won't opt for laptop specially U-series processor. so that many thread will in all possibilities not help the laptop.
1. light duty software cares if there is 4ghz or 4,5ghz?
2. intel disagreed with you. they are produses up to 8 cores in mobile.
3. higher frequencies-less energy efficient.
4.and finally having extra cores will not hurt you. light-threaded? extra frequency to a couple of cores, idle for the other.

Quotethirdly going forward intel is improving on graphics like anything.
who cares? do you know a successful integrated graphics which grabbed a market share or made some crazy earnings? igpu is by far the less profitable waste of silicon. half of the chip-200usd cpu, half-20usd igpu. you can always counter any integrated with discrete. even with the same piece of silicon, just adding faster memory and higher frequencies.
and..why do people misses the point about shared power limit? tiger lake is crazy fast and his gpu is crazy fast but both 15w total. how is that possible? in reality-zen 2 is more efficient than intel offerings, and amd or nvidia graphics are more efficient than intel offerings. so where is the trick?
QuoteBut IPC advantage of intel is still main concern
maybe. or mabe not. yep, ice lake is pretty fast. tiger would be 10% faster, due to 10% higher frequencies. and now tell me why there is no gaming or working machine with ice lake?
Quoteintel can compete better even if mfg tech is one step ahead like in current case of TSMC vs intel, 10gen intel is still competitive.
10 and 7 are marketing names. pretty safe to say that intel's 10nm= tsmc's 7nm https://fuse.wikichip.org/wp-content/uploads/2019/06/tsmc-intel-xtor-comp-16-10-7.png
also intel's 14nm was clearly better than tsmc's 16 or samsung/glofo 14.

I think on a CPU level, things are leveling out and AMD has to be commended here. But I am still wary about what kind of laptops OEMs actually push out. Intel's hegemony can't be discounted all of a sudden. We did see some cool designs at CES and about 100 designs are on the anvil, but for a common person, it is still Intel Inside, at least when it comes to laptops or ultrabooks.
Posted by william blake
 - January 27, 2020, 04:51:45
Quote from: k on January 27, 2020, 03:52:20
firstly, with better brand name and deeper distribution chain, intel is always with premium configuration.
kinda common knowledge. but not about technology or products.
Quotesecondly, most light duty software are not mutlithread ready and heavy one's users won't opt for laptop specially U-series processor. so that many thread will in all possibilities not help the laptop.
1. light duty software cares if there is 4ghz or 4,5ghz?
2. intel disagreed with you. they are produses up to 8 cores in mobile.
3. higher frequencies-less energy efficient.
4.and finally having extra cores will not hurt you. light-threaded? extra frequency to a couple of cores, idle for the other.

Quotethirdly going forward intel is improving on graphics like anything.
who cares? do you know a successful integrated graphics which grabbed a market share or made some crazy earnings? igpu is by far the less profitable waste of silicon. half of the chip-200usd cpu, half-20usd igpu. you can always counter any integrated with discrete. even with the same piece of silicon, just adding faster memory and higher frequencies.
and..why do people misses the point about shared power limit? tiger lake is crazy fast and his gpu is crazy fast but both 15w total. how is that possible? in reality-zen 2 is more efficient than intel offerings, and amd or nvidia graphics are more efficient than intel offerings. so where is the trick?
QuoteBut IPC advantage of intel is still main concern
maybe. or mabe not. yep, ice lake is pretty fast. tiger would be 10% faster, due to 10% higher frequencies. and now tell me why there is no gaming or working machine with ice lake?
Quoteintel can compete better even if mfg tech is one step ahead like in current case of TSMC vs intel, 10gen intel is still competitive.
10 and 7 are marketing names. pretty safe to say that intel's 10nm= tsmc's 7nm https://fuse.wikichip.org/wp-content/uploads/2019/06/tsmc-intel-xtor-comp-16-10-7.png
also intel's 14nm was clearly better than tsmc's 16 or samsung/glofo 14.
Posted by k
 - January 27, 2020, 03:52:20
actually for desktop your people arguments may be true like not tweaked single channel memory and bla-bla. but you are talking about laptop. here are difference.
firstly, with better brand name and deeper distribution chain, intel is always with premium configuration. like it quoted in CES as well, 4850h has no support beyond 2060 graphics. this is because OEM choke AMD as they target value market with that. that way actual configurations specially with U-series processor where customer is not looking for performance as primary, in all possibilities will be like one used by author. And you are helpless as you cannot assemble a laptop.
secondly, most light duty software are not mutlithread ready and heavy one's users won't opt for laptop specially U-series processor. so that many thread will in all possibilities not help the laptop.
thirdly going forward intel is improving on graphics like anything, and AMD with worry for discrete market is choking intentionally their APUs. First time i am surprised Intel at par or better with AMD igpu. you are right Quad socket and Quad SMT of zen3 will make things look good for AMD on server but never expect that on laptop. Finally intel is against its habbit looking very serious on technology this time. they hired jim keller and raja. So USP of AMD like infinity fabric and MCM won't help too long. As their developer are already with intel and will make atleast at par if not better. Only advantage AMD is left with is Agility with manufacturing they can switch to best whether TSMC or Samsung or even intel very quickly. but intel needs to fight with everyone if they want to survive. But IPC advantage of intel is still main concern. go back into history GF was always lagging behind intel but AMD ruled because of better IPC even with lagging Mfg tech. Same way with better IPC intel can compete better even if mfg tech is one step ahead like in current case of TSMC vs intel, 10gen intel is still competitive.
Posted by william blake
 - January 27, 2020, 01:05:37
Quote from: deksman2 on January 26, 2020, 23:47:21
If those results are actually accurate and not fake... then the 4800U is an excellent competitor for both those Intel chips.
It offers similar enough single threaded performance, coupled with better multithreaded performance in a 15W TDP with 8 cores/16 threads and more powerful graphics.
Given the efficiency of Zen 2 in general along with its availability (which is likely going it be sparse or close to non-existent for Intel due to their 10nm woes), AMD still (in all likelihood) has the advantage.
Also, the fact AMD was able to cram 8 cores and 16 threads into a 15W TDP (which can be configured up to 28W actually and these tests likely show only 15W limit) whereas Intel couldn't, and that it generally stays within its TDP limits while Intel tends to go WAY overboard, I don't think Intel has good competitor here.
And by the time Tiger Lake comes out, it will still have to deal with Zen 3.
pretty much this. intel needs 3 lineups to counter a single chip :)
people totally missed that.
look at the difference between 15 and 45 watt renoirs. same turbo. same gpu. they not even tried. zero extra work. no tuning with different power/frequency ladder, but the day they announced it i cancelled ice lake laptop preorder for my colleague and told him to wait if his needs are video and picture editing in a thin form factor. because they made exactly i(and market?) wanted. more cores less gpu and no extra chiplet. fairly small chip, no high(and fake) frequencies for gaming variant. all that meant their goal was efficiency and sheer might. and we all saw what is zen2 is capable of doing in this field already.
Su also was very confident. and oems too. and another thing to consider-the future ice lake server, core count and power draw, will be way less efficient than current zen 2. so with all that im pretty sure all renoirs are very competitive, from top to bottom.
another, but small, thing-new 2 core chip for extra cheap. no more bulldozer crap in the low end, that was really a crime against consumers. so, 2019 and 2020 is night and day for amd mobile.
you kinda wrong about tiger lake i think, its coming in summer afair, zen 3 apu a year after. it should be another big jump, new cpu arch+rdna 2, but we have a whole year for tiger vs renoir battle.
Posted by heffeque
 - January 27, 2020, 00:38:05
I'm all for AMD in desktop and server, but sadly 3X00U is way behind Intel in performance and especially in idle power consumption (see Surface Pro 7 Intel vs AMD comparison), and I highly doubt that in 4X00U both performance and idle consumption will have a huge huge huuuge bump to put it in front of Intel.

I hope I'm wrong.
Posted by deksman2
 - January 26, 2020, 23:47:21
Quote
should be? they took the only renoir result available for geekbench.
ok, here is 4800u with dual channel versus peak(!) results from 1065g7 and 10710u
.
.
.
ice lake with the fastest memory possible, 3733
comet lake with fastest for him 2666, 4+ghz avg(be sure its not 15w)
renoir with 3200, all dual channels.

If those results are actually accurate and not fake... then the 4800U is an excellent competitor for both those Intel chips.

It offers similar enough single threaded performance, coupled with better multithreaded performance in a 15W TDP with 8 cores/16 threads and more powerful graphics.

Given the efficiency of Zen 2 in general along with its availability (which is likely going it be sparse or close to non-existent for Intel due to their 10nm woes), AMD still (in all likelihood) has the advantage.

Also, the fact AMD was able to cram 8 cores and 16 threads into a 15W TDP (which can be configured up to 28W actually and these tests likely show only 15W limit) whereas Intel couldn't, and that it generally stays within its TDP limits while Intel tends to go WAY overboard, I don't think Intel has good competitor here.

And by the time Tiger Lake comes out, it will still have to deal with Zen 3.

Of course, I could be wrong, but we shall see.
Posted by william blake
 - January 26, 2020, 23:13:14
Quote from: Ed on January 26, 2020, 21:55:48
Isn't this comparison fairly pointless?

Using Intel Top Model's number and Compared to 4700U when it should be 4800U. Using Single Channels which will lower pretty much all results including CPU , Single, Multi and GPU benchmarks.
should be? they took the only renoir result available for geekbench.
ok, here is 4800u with dual channel versus peak(!) results from 1065g7 and 10710u
https://www.userbenchmark.com/UserRun/23316225
https://www.userbenchmark.com/UserRun/23909249
https://www.userbenchmark.com/UserRun/21896092
ice lake with the fastest memory possible, 3733
comet lake with fastest for him 2666, 4+ghz avg(be sure its not 15w)
renoir with 3200, all dual channels.
Posted by Ed
 - January 26, 2020, 21:55:48
Isn't this comparison fairly pointless?

Using Intel Top Model's number and Compared to 4700U when it should be 4800U. Using Single Channels which will lower pretty much all results including CPU , Single, Multi and GPU benchmarks.

Then those Intel CPU's PL2 aren't even similar to AMD's PL2.
Posted by pro at cpu
 - January 26, 2020, 20:47:37
Intel die size wont do it forget about 5ghz it will never reach there in long duration. I bet the selling power of their laptop is marketing all of their laptop 5ghz. Why would they do this because higher number is better and when there is two cpu companies the market are split. Intel cannot take all and everything and one day they will lose it.