Ryzen 2500u benchmark by The Tech Report (img heavy)

[AlTech]

2 minutes ago, Prysin said:

been 2-3 leaks out already... you can find them on google

Just leaks? Nothing actually real or tangible to go on?

 

Oh ok. Good to know. Cos reading your post it felt like you had like 5 Ryzen laptops hooked up on a Battery test or something.

 

I'll wait for reviews from techie youtubers before coming to a conclusion on battery life and GPU performance.

 

I know you're not the biggest AMD fan ever but I don't like you trying to give people a bad impression and/or trying to influence people's opinions like that.

[AlTech]

1 minute ago, gabrielcarvfer said:

Seems pretty terrible.
https://hothardware.com/reviews/ryzen-mobile-benchmarks-and-performance-analysis?page=3

The testing methodology used in that review for battery life is deeply flawed.

 

A video loop test is nothing short of completely not representing real world usage. Same for stuff like PCMark's battery life test.

 

Just using the device over the course of day doing what the reviewer normally does on their laptops and repeating this several times should be just fine.

[NumLock21]

12 minutes ago, Prysin said:

if you look at the current prices. we see that AMD isnt CLOSE to competitive.... it starts at 700+ if you want a 256gb SSD...

It's better to get the HDD one and then add a M.2 SATA or NVMe ssd yourself. The SSD sku don't have the required cables and mounting for the HDD. Those cost extra, from HP, not including shipping and tax comes out to $42 USD. Also the memory is under that shiny cover, that cover itself cost $22 USD.

 

8 minutes ago, Trixanity said:

Interesting that Notebookcheck had pretty much the opposite conclusions from their benchmarks. Guess we need more sources and quite frankly more devices to test with.

Anandtech will also be getting one for review.

[AlTech]

2 minutes ago, gabrielcarvfer said:

Meh, video looping is the standard of the industry. The NotebookCheck review show much better results in terms of power draw on different scenarios, bot not total time of usage.

And using processed meat is the Industry Standard way for Fast Food companies to make burgers. Does that mean that it's necessarily good?

 

The video loop test results could be caused by anything from a lack of a dedicated VP8/VP9 codec (i don't know or have any information on this, I'm purely speculating) all the way to an inefficient architecture overall (which we're pretty sure we know isn't true from Ryzen Desktop).

[Prysin]

9 minutes ago, AluminiumTech said:

Just leaks? Nothing actually real or tangible to go on?

 

Oh ok. Good to know. Cos reading your post it felt like you had like 5 Ryzen laptops hooked up on a Battery test or something.

 

I'll wait for reviews from techie youtubers before coming to a conclusion on battery life and GPU performance.

 

I know you're not the biggest AMD fan ever but I don't like you trying to give people a bad impression and/or trying to influence people's opinions like that.

heard of NDA??? Its what reviewers have to sign up for if they want review samples..

 

Guess where leaks come from? Reviewers that dont care about the NDA

[AlTech]

Just now, Prysin said:

Guess where leaks come from?

No clue. But i'm pretty sure they wouldn't come from:
 

Quote

Reviewers that dont care about the NDA

 

Leaks can be wildly inaccurate and generally I don't trust them too much.

[NumLock21]

Intel with MX150. CPU on the left and GPU on the right

 

AMD Ryzen with VEGA 8. Intel ones with onboard graphics has the same type of cooling as this one.

 

RAM is actually under the silver cover next to the M.2 slot

 

Sources:

1st pic, from me

2nd and 3rd pic from

http://apusilicon.com/

[Taf the Ghost]

One problem AMD has when getting back into actual competition in a lot of these spaces is the classic "Well, Internet Explorer works this way" issue. When all of the testing is done on Intel systems for comparing Intel systems, you don't know if your testing is will be as accurate when comparing it to a similar product that's not from Intel.

 

Though I suspect the VLC Loop testing has more to do with Raja's inability to get good Drivers out on time than the entire Raven Ridge design. (Though I really don't get what is up with that memory latency result. How is it worse than Ryzen with 1 CCX?)

 

If there's a host of Vega driver issues, it would explain the rather varied results we've gotten already for power usage. As for gaming, it's literally at almost exactly the result we could predict. It's weaker than the MX150 but better than the Intel HD graphics. You can play Esports titles without much issue at an acceptable speed, which is why I think the desktop APUs will be a really good seller when those land.

[NumLock21]

It also support Free Sync

[Trixanity]

55 minutes ago, NumLock21 said:

Anandtech will also be getting one for review.

That means we'll get that review in about three months.

 

49 minutes ago, Prysin said:

heard of NDA??? Its what reviewers have to sign up for if they want review samples..

 

Guess where leaks come from? Reviewers that dont care about the NDA

NDA? On a device you can pick up from the store right now? 

 

These aren't leaks. They're preliminary testing.

[Cookybiscuit]

AMD finished and bankrupt, they can't recover from this one.

[Shadestones]

I'm hoping future iterations of mobile Ryzen perform better. It would be great for AMD if they could compete with Intel in that space.

 

I've only seen AMD chips in the past on severely underpowered laptops.

[JediFragger]

So, Intel's share price is gonna get another bump then...

[WereCat]

2 hours ago, Prysin said:

for gaming, it probably cost them 8 FPS, maybe 12. But thats it. Its still slower then a MX150

I mean the frametimes and memory access latency.

I have never seen so bad latency and I am convinced that this is the cause of the seemingly underwhelming results.

[WMGroomAK]

23 minutes ago, WereCat said:

I mean the frametimes and memory access latency.

I have never seen so bad latency and I am convinced that this is the cause of the seemingly underwhelming results.

I think one of my questions concerning the latency issue would be whether it is limited as well due to the DDR4 speeds as that appears to have an impact on the Infinity Fabric performance.  

[Daniel644]

2 hours ago, AluminiumTech said:

The testing methodology used in that review for battery life is deeply flawed.

 

A video loop test is nothing short of completely not representing real world usage. Same for stuff like PCMark's battery life test.

 

Just using the device over the course of day doing what the reviewer normally does on their laptops and repeating this several times should be just fine.

flawed or not the worst case is it doesn't give you the accurate how long it will last, using a test like that on everything gives you comparative data, like this unit lasted x amount of time and unit from last week lasted y amount of time, x is 20% longer so laptop x has 20% better battery life, better method then random usage because everybodies usage varies (example I have an old Galaxy Tab 7" that I use STRICTLY for music playback in the car, I get 40+ DAYS per charge, where if I use the screen to watch video I might get a few hours to better part of a day out of it), trying to state how long the battery will last for your use is a flawed concept, better inaccurate total time numbers that give comparative data.

[Nup]

Solidified vegas shortcomings? I would have bet that with a situation where it could be tuned down a notch it would come into its prime. A bit of shame. 

Maybe drivers will help with all this. 

[NumLock21]

Looks like some buyers just look at the price and not the specs.

 

[thorhammerz]

23 minutes ago, NumLock21 said:

Looks like some buyers just look at the price and not the specs.

That's actually a considerable segment of the market.

 

Which prebuilts can do (at a minimum) X? Of those machines, which one is the cheapest? Sold!! 

[Okjoek]

I want to see more sources and tested devices.

[Misanthrope]

8 hours ago, Prysin said:

for gaming, it probably cost them 8 FPS, maybe 12. But thats it. Its still slower then a MX150

Not for an APU that lives and dies on ram speeds

[Prysin]

5 hours ago, Misanthrope said:

Not for an APU that lives and dies on ram speeds

i have toyed around a lot with APUs, got several of them. It REALLY depends, but generally you see a 8-10 FPS uplift.

 

Now, after looking over HP's site, it appears that they only provide some form of dual channel (2x4GB SODIMM, 1x4GB soldered + 1x8GB SODIMM or 2x 8GB SODIMM)

 

 

[Prysin]

12 hours ago, AluminiumTech said:

No clue. But i'm pretty sure they wouldn't come from:
 

 

Leaks can be wildly inaccurate and generally I don't trust them too much.

Its all about the source.... Leaks coming from CNN, maybe not worth a shit... Leaks coming from Wikileaks or Snowden - well, they have yet to be proven wrong. May not like their leaks, but they are factual.

 

So again, all depends on the source... Many tech tubers and review sites are lazy, they dont test things properly. They do half baked tests, relying on automation without overseeing the process. That means any misleading optimization on the HW manufacturers side will go unnoticed for a long long while

[MyName13]

When are these coming out?

[Prysin]

20 minutes ago, MyName13 said:

When are these coming out?

they are out, or atleast for pre.order