4900HS - First benchmarks at Hardwareunboxed

[LukeSavenije]

Introduction:

The first benchmarks from hardware unboxed are now out, and 4900HS tops almost every chart against it's competitors, giving AMD a whole new lead in performance

 

Cinebench R20

Spoiler

 

Cinebench R15:

Spoiler

 

Handbrake x265:

Spoiler

 

Blender:

Spoiler

 

7-zip

Spoiler

 

Photoshop:

Spoiler

 

PCmark 10:

Spoiler

 

Excel:

Spoiler

 

Matlab:

Spoiler

 

Adobe PDF export:

Spoiler

 

Sisoft Sandra AES-256:

Spoiler

 

Adobe Premiere:

Spoiler

 

GTA V:

Spoiler

 

Civilasation VI:

Spoiler

 

CS:GO

Spoiler

 

Gears 5:

Spoiler

 

Sisoft Sandra (memory latency)

Spoiler

 

Clockspeed and power (handbrake)

Spoiler

 

4900HS vs 9880H

Spoiler

 

4900HS vs 9880H

Spoiler

 

4900HS vs 9750H

Spoiler

 

First reaction:

insane... outright insane. I hope very much to see more of AMD like this in the future, because the have shown to undercut intel in various tasks with a lower power consumption. May there be competition!

 

Source:

 

[greenhorn]

Here is an additional source for some performance figures:

It looks impressive

[Ashley MLP Fangirl]

i can't wait for an 8 core 16 thread iMac or a 13" 6 core MacBook Pro... 

 

if only Apple used Ryzen... 

[Tristerin]

My laptop I use for productivity was expensive AF.  And just checking the numbers (I crunch huge excel calcs) That should beat out my Xeon E2276M in this process which is only slightly faster (real time) than a i7 9850H.  Just beautiful.  I could see a laptop with this Ryzen CPU like this being thousands less than my laptop with the rest of the specs being same.

[porina]

From when I did limited (low) power testing on desktop Zen 2, the scaling looked very promising and I looked forward to Zen 2 based mobile solutions. At first glance, this doesn't disappoint, but will need going over in more detail later. While it comfortably beats the Intel, we have to bare in mind the CPU it went against is still essentially a Skylake derivation so wont have the new architecture. Then again, I don't think they offer the new architecture in 8 cores yet, so there isn't anything else better to compare against yet.

 

While many here thought the AMD attack on desktop was hurting Intel, the mobile space I think is where we'll see the mainstream damage done. This was the last piece of the pie Intel had to itself. 

[Curufinwe_wins]

6 minutes ago, porina said:

From when I did limited (low) power testing on desktop Zen 2, the scaling looked very promising and I looked forward to Zen 2 based mobile solutions. At first glance, this doesn't disappoint, but will need going over in more detail later. While it comfortably beats the Intel, we have to bare in mind the CPU it went against is still essentially a Skylake derivation so wont have the new architecture. Then again, I don't think they offer the new architecture in 8 cores yet, so there isn't anything else better to compare against yet.

 

While many here thought the AMD attack on desktop was hurting Intel, the mobile space I think is where we'll see the mainstream damage done. This was the last piece of the pie Intel had to itself. 

For me, the real kicker is if AMD can get their -U equivalent processors competitive at idle/low power usage. The upper end of mobile cpu's is effectively the same situation as desktops IMO, and the thin'n'lights are where the real money is made.

 

I was sadly disappointed with how the Surface Laptop Zen+ processors performed in that metric.

 

 

This is the single most important thing that matters in today's market.... outside of gaming laptops that is.

 

[5x5]

The king is dead, a new king has Ryzen

[porina]

6 minutes ago, Curufinwe_wins said:

I was sadly disappointed with how the Surface Laptop Zen+ processors performed in that metric.

Zen+ - there's your problem. Zen 2 is where the fun starts.

[Curufinwe_wins]

8 minutes ago, porina said:

Zen+ - there's your problem. Zen 2 is where the fun starts.

Zen2 doesn't mean shit if the idle/video load power isn't also dropped down to be competitive with Intel's current offerings, and that hasn't been tested yet afaik. That was all I was saying.

 

I hope they get it down, but point is the high-level operation stuff is far less relevant to 95% of the market than the idle/low load usage capabilities.

 

If they get it down, then these chips will be gods amongst men. If not, it's almost straight up DOA.

[RejZoR]

That's weird. I just benchmarked my Ryzen 5 2500U in R20 and it scored 1386 points. Which puts it above Ryzen 7 3700U. How is that possible?

 

My laptop is a rather generic HP laptop from their 17-something lineup. Ryzen 2500U, 8GB RAM in dual channel and 256GB DRAMless NVMe SSD. Only thing special about it is that I replaced factory thermal paste with CoolerMaster Maker Nano when I was cleaning it last time. I don't think thermal paste would make such an improvement that 2500U would beat 3700U chip... So, something is a bit weird with the scores or something on the graph...

[SpaceGhostC2C]

37 minutes ago, VegetableStu said:

"Battery life will also be on people’s minds, but it’s not something we set out to test as we didn't have enough data for various laptops to make it a fair comparison."

Well, battery life will depend on the specific laptop, but performance per watt is the relevant metric here, I think, rather than raw benchmarks. Laptop CPU can't possibly give you the most performance, their whole point is to perform "within the constraints of a portable device", so the key is how they strike that balance (also, "high-end gaming laptops" are stupid ). I mean, there's even a 90W configuration of a "laptop CPU" in OP's charts. What's even the point in that? It's going to be decimated by a number of 65W desktop CPUs anyway...

 

42 minutes ago, VegetableStu said:

"Even more astonishing is AMD’s ability to offer so much performance in such tight thermal and power constraints."

Not that is what ultimately matters for a laptop part, I think: the ability to choose a lot of battery time, or higher performance while remaining a mobile device.

[Curufinwe_wins]

1 minute ago, RejZoR said:

That's weird. I just benchmarked my Ryzen 5 2500U in R20 and it scored 1386 points. Which puts it above Ryzen 7 3700U. How is that possible?

 

My laptop is a rather generic HP laptop from their 17-something lineup. Ryzen 2500U, 8GB RAM in dual channel and 256GB DRAMless NVMe SSD. Only thing special about it is that I replaced factory thermal paste with CoolerMaster Maker Nano when I was cleaning it last time. I don't think thermal paste would make such an improvement that 2500U would beat 3700U chip... So, something is a bit weird with the scores or something on the graph...

up-tdp may be configured differently. or ram configuration might be different. It's really hard to say, and honestly I don't have a ton of trust in the rigor of HUB, but we will see with more details moving forward.

[williamcll]

Going to be interesting to see what intel has to offer next week.

[Curufinwe_wins]

13 minutes ago, SpaceGhostC2C said:

Well, battery life will depend on the specific laptop, but performance per watt is the relevant metric here...

 

Not that is what ultimately matters for a laptop part, I think: the ability to choose a lot of battery time, or higher performance while remaining a mobile device.

Even that isn't relevant IMO. Battery life is dependent on specific laptop, but places like Anandtech do a good job at trying to isolate down to just processor differences, and it is the idle/low load (or hardware accelerated video load) that really matters for battery life specs being thrown out.

 

Which yes is variable, and hard to measure, but it is the real thing that matters for company marketing and even most consumer usage patterns.

 

 

For everyone's record, I'm not trying to shit on Zen2 or make a moving goalpost or anything like that. I'm more just cautioning against saying the sky is falling for Intel until we know if they can actually deliver the goods that really matters in the market.

[CarlBar]

17 minutes ago, Curufinwe_wins said:

up-tdp may be configured differently. or ram configuration might be different. It's really hard to say, and honestly I don't have a ton of trust in the rigor of HUB, but we will see with more details moving forward.

 

Given GN and LTT have both used them to cross check unexpected results, (and vice versa), that's not a good assumption to make IMO.

 

 

Also people prophesying intels doom and gloom need to remember. TSMC only has so much 7nm fab capacity. Even if on the back of this everyone who could went Zen2 Intel would still have an overwhelming percentage of the market share because AMD simply can't get enough silicon to take a significant minority, let alone a majority, of the market share.

[Curufinwe_wins]

23 minutes ago, CarlBar said:

 

Given GN and LTT have both used them to cross check unexpected results, (and vice versa), that's not a good assumption to make IMO.

 

 

Also people prophesying intels doom and gloom need to remember. TSMC only has so much 7nm fab capacity. Even if on the back of this everyone who could went Zen2 Intel would still have an overwhelming percentage of the market share because AMD simply can't get enough silicon to take a significant minority, let alone a majority, of the market share.

Both have also identified huge issues in their results (and exceedingly biased world view conclusions) in the past as well.

 

It just isn't a source I would first-view take as god's word or anything close to honestly. I wouldn't come close to trusting LTT for that either (LTT's rigor is a joke these days honestly).

 

But your other point is also valid. People look at intel saying they don't have capacity to keep stuff in stock as indication that production is still super stifled. That is incorrect, demand had just been THAT high.

[JoostinOnline]

It's really weird seeing Ryzen with a 4000 series, because I keep thinking it's Haswell architecture. Of course, I still have a 4790k.

[5x5]

Interesting - Dave2D is claiming 8 hours of battery but NBC are claiming 4. Could be the 4800HS is MUCH more efficient since he's testing that? Or the BIOS is still in beta.

 

[Fasterthannothing]

Gosh darn it AMD this is unbelievable I want AMD in everything! The only problem is AMD keeps making bigger performance improvements each generation. Making me feel the need to buy a new gen every time because the performance per $ is so darn high.

[Curufinwe_wins]

8 minutes ago, 5x5 said:

Interesting - Dave2D is claiming 8 hours of battery but NBC are claiming 4. Could be the 4800HS is MUCH more efficient since he's testing that? Or the BIOS is still in beta.

 

maybe beta bios, but more likely different workloads having huge performance impacts. Wouldn't be that surprising. PCMark battery life for example is very different than netflix battery life.

[Doobeedoo]

That is really very impressive across the board though. 

[genexis_x]

AMD YES

[Taf the Ghost]

3 hours ago, porina said:

Zen+ - there's your problem. Zen 2 is where the fun starts.

 

3 hours ago, Curufinwe_wins said:

Zen2 doesn't mean shit if the idle/video load power isn't also dropped down to be competitive with Intel's current offerings, and that hasn't been tested yet afaik. That was all I was saying.

 

I hope they get it down, but point is the high-level operation stuff is far less relevant to 95% of the market than the idle/low load usage capabilities.

 

If they get it down, then these chips will be gods amongst men. If not, it's almost straight up DOA.

AMD was apparently rather upset with Microsoft in how they handled a lot of the Surface stuff. While a big design win for AMD, the work into the power control and other parts was a tad lacking. For that low of idle state and low power state processing, it's down to firmware and platform development, so AMD still has a ways to go in their relationships, in taht regard.

[Taf the Ghost]

On the 4000 series APUs, in general, they actually came out better than expected. Some of these results were always going to happen, as we know what 8c zen2 looks like in power constrained situations. But, AMD's engineers still found a lot with this generation, and now things get fun.

[Curufinwe_wins]

14 minutes ago, Taf the Ghost said:

 

AMD was apparently rather upset with Microsoft in how they handled a lot of the Surface stuff. While a big design win for AMD, the work into the power control and other parts was a tad lacking. For that low of idle state and low power state processing, it's down to firmware and platform development, so AMD still has a ways to go in their relationships, in taht regard.

Fair, but it isn't like Microsoft is likely to have put less effort into it than they did with intel's platform (unless ofc Intel did more of the work for microsoft there, which admittedly is VERY possible, just as Nvidia is known to give massive developer resources to it's partnered games).

 

BTW, looks like the Zeph G14 takes the Surface Book 2 approach to battery life. Which don't get me wrong, I 100% approve of the technique (I own a SB2 afterall haha), but LTT in general praising the G14's 72Whr battery life compared to a higher brightness, higher resolution, touchscreen 51Whr XPS13 is obviously not apples to apples by any stretch. 

 

I'm super interested to see what happens given the same platform (say if Dell made a Zen 2 XPS 15).