AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

for desktop use interested in 200% scaling and the added color range etc..

 

I have a 4K 17" laptop. I really like it; text is much crisper and I can use a dimensionally smaller but more pixel font to get more on the screen, which I like for software development. It's great for image editing. I don't game beyond Minesweeper on the lowest level.

I'm running Linux.

EDIT: Another advantage of the 4K screen is that when I connect it to an external monitor all of my windows and icons have the same position and pixel size on either screen.

 

That's cool, still, we're back to square 1 of the fact that most OEM's don't offer laptops with quality 4K panels, and if some do, you are usually stuck with the config they offer (not much choice in the matter).

 

Workstation laptops (Lenovo P, Dell Precision, HP Z-book) tend to offer pretty good options, and memory and storage are field replaceable. Be nice to have some of the Ryzen 2 chips as options, although by the time I replace my current one, that would probably be Ryzen 4 or so.

 

You mean Ryzen 3 (as in Zen 2).
AMD options are pretty scarce to non-existent in this arena (at least going beyond 8 cores).
The only all AMD options that became available in the last few years were Asus GL702ZC (one I recommend to avoid like mad), and Acer Predator Helios 500 PH517-61 (highly recommended).

By workstation laptops... I usually count desktop replacements mainly with high number of cores (Zen 2 mobile muddies that somewhat as you technically don't need an 8c/16th desktop CPU running in a laptop since the mobile version runs very close to it and comes with an iGP at less TDP).

Though Zen 2 brought desktop power into a mobile format, so there's less of a need to put a real desktop processor into a laptop (unless you count Clevo of course which goes beyond 8 cores, but their 15" option of Zen 2 desktop CPU's implementation is running too hot and loud, so it would be even worse than Asus GL702ZC in that regard, at least right now - they have a long way to modify the fans and optimize the hw to reduce noise - but putting all that into 15" chassis was a BAD call).

I guess today you can pick and choose mostly from 4800H and 4900H... heck, 4800u is easily on the level of my 2700 cpu-wise, but it might result in better lower minimums and average frame rates if paired with a discrete GPU.

 

Threadripper's compatibility screw up once again, but this time for mainstream.

AMD B550 Chipset Detailed, It's Ready for Zen 3, Older AM4 Motherboards not Compatible Techpowerup.com | Today

In their briefing leading up to today's Ryzen 3 3100 and 3300X review embargo, AMD disclosed that its upcoming "Zen 3" 4th generation Ryzen desktop processors will only support AMD 500-series (or later) chipsets. The next-gen processors will not work with older 400-series or 300-series chipsets. This comes as a blow to those who bought premium X470 motherboards hoping for latest CPU compatibility running into 2020.

 

This is highly unusual.
In that article, it says that AMD ran into ROM size issues for fitting support for all the older CPU's into the microcode.

I find that a bit strange since a lot of B350/X370 and B450/X470 mobos for example had more than large enough ROM chips to fit the microcodes (obviously, some did not, but that was the OEM fault, not AMD's).

In fact, previous 'lack of support' was on OEM's not fitting large enough ROM chips and OEM's reluctance in releasing a BIOS update (not AMD) for specific mobos.

In case of Acer in my Predator Helios 500, the ROM size is (I think) 4 times larger than what is needed for Zen 2 support... but Acer never released BIOS updates for Zen 2 (not because of ROM sizes but because they simply stopped supporting the unit).

I find it unlikely that similar desktop B450/X470 mobos wouldn't have the ROM capacity to house Zen 3 support.

If it does come down to ROM chip sizes again, why don't they simply write out certain CPU's (such as lower end first gen Zen or just the entire Zen 1 generation) out of the microcode if they're concerned or leave it up to motherboard manufacturers to remove older CPU's or leave them in if the ROM chip size is large enough).

EDIT: It won't matter for me obviously since Acer never bothered with updating the BIOS anyway, but for desktop users using B450/X470 who wanted to upgrade to Zen 3... its a big let-down.

Plus, there are details like ROM chip sizes which aren't included in mobo specs really (at least not on the technical specs page of the product in question).

 

The Exciting Future of AMD Socket AM4 community.amd.com

Q: What about (X pre-500 Series chipset)?
A: AMD has no plans to introduce “Zen 3” architecture support for older chipsets. While we wish could enable full support for every processor on every chipset, the flash memory chips that store BIOS settings and support have capacity limitations. Given these limitations, and the unprecedented longevity of the AM4 socket, there will inevitably be a time and place where a transition to free up space is necessary—the AMD 500 Series chipsets are that time.

Robert Hallock is the Technical Marketing lead for AMD consumer processors. The information contained herein is for informational purposes only and is subject to change without notice. Timelines, roadmaps, and/or product release dates are plans only and subject to change. “Zen"; "Zen+"; “Zen 2”; and “Zen 3” are codenames for AMD architectures and are not product names. GD-122

 

I forewarned, only build for today! I was just considered being a troll, told ya all.

Edit; now that they are top dog performance wise they no longer care about enthusiasts looking at upgrade paths and can do what ever they want!

 

You are WRONG! This problem happened with Zen+ or Zen 2, which is why some boards needed a complete switch of bios because it couldn't fit the microcode for multiple gens with new features, etc. Look up the old bios chip size controversy. This is not the first time amd used the excuse. Just cannot remember if it was 8MB or 16MB bioses that were the issue.

Edit: also, you cannot blame mfrs. When this is a repeat. At some point, amd should have required larger chips. Seems the x570 and b550 have that or rather it has it by not supporting first gen chips.

Sent from my SM-G975U1 using Tapatalk

Well, we botched and moaned about no warning. Let's give credit that they are telling consumers 4-5 months before launch so they know how much to save up. They learned something from our complaining and not buying the new platform. Instead, let the butt hurt happen now and that way they will still buy this fall.

Too bad our economy won't be that recovered by then (not talking politics, just unemployment numbers)...

Sent from my SM-G975U1 using Tapatalk

 

If you recheck my reply, you will see near the very beginning I DID in fact mention that 'many' mobos had sufficiently large ROM chips to fit microcode updates (which did receive the updateS) but that obviously, 'some' did not (Which was in effect a problem with OEM's not fitting large enough ROM chips - some users here suggested the OEM's were too busy putting flashy graphics into the UEFI which took up space).

MFRS?
Not familiar with the acronym.

The X570 has been out in the wild for a long time now and B550 is fairly new.
AMD mentioned well before this that AM4 will indeed be compatible with Zen 3... however, we (the users) indeed thought that this would extend to existing X370, B450/X470 mobo's with large enough ROM chips.

It would be noteworthy to check just how large the ROM chips on the B550 and X570 mobos are and compare them to B450 and X470 mobos ROM chip sizes and see if AMD simply decided to cut out B450/X470 and earlier mobos for purely monetary reasons or if the ROM chips on B550 and X570 are indeed larger.

Here we go:
https://www.techpowerup.com/257201/...ail-amds-zen2-backwards-compatibility-promise

This is basically saying the OEM's were using too small ROM chips and had to cut back on UI graphics to make room for microcode updates due to using 16MB ROM chips

Now with Zen 3, its AMD itself saying that they decided to cut out support on mobos that aren't B550/X570 due to ROM chip limitations (which would make sense for mobos that have insufficient ROM chip sizes, but as I noted in my reply, I doubt this would apply to ALL mobos as not all of them came with underwhelming ROM chip sizes).

If we know that OEM's CAN fit Zen 2 microcode updates into 16MB/32MB ROM chips... then I don't think a 64MB/128MB ROM chip would be 'too small' for Zen 3.

My B450 comes with 128MB ROM chip, and its a desktop replacement. I'm pretty sure that manufacturers producing quality mobos would give their full blown desktop mobos 128Mb/256MB ROM chip sizes (or at least a good chunk of them would).
I know that most mainstream mobos came with 16MB - which is barely enough if you cut out the fancy UI... but realistically, there had to be mobos on the market with 32/64/128 MB ROM chips too.

Here's the Google DOC comparing BIOS ROM chip sizes:
https://docs.google.com/spreadsheet...FnsZYZiW1pfiDZnKCjaXyzd1o/edit#gid=2112472504

Looks like a fair number of X370, B450, X470 and even X570 mobos came with 16MB and 32MB ROM chips at least (most X370 seem to have only 16MB though).

So, yes, ROM chip size would make a difference, but only on mobo's that don't have large enough ROM chips... and as we know from before, by trimming the UEFI graphics, mobo manuf. were able to fit Zen 2 microcode updates onto the 16MB ROM chips.

A few of X570 Mobos in that google DOC seem to come with 16MB ROM chips too and AMD said they will support ALL B550 and X570 mobos (but no previous models).

Well, it seems the ROM chip size is not a good argument on AMD's side... but as I said, OEM's were also to blame (before) for giving various mobos small ROM chips and then filling them with fancy UI graphics that took up needless space.

Still, I don't think this is a good 'excuse' for AMD.

 

The Predator Helios 500 AMD only has a 16MBbyte (128Mbit) ROM chip. MXIC MX25U12835F. It is about 47% utilized. I don't have a dump of the NH58AF1 B450 system yet, but I could guess it has a higher utilization percent for the newer processor microcode in AGESA, and presence of a UEFI graphical interface. Honestly I would not mind dropping the "graphical" UI if it meant space for newer processor microcode, but it seems like Clevo has standardized it across all of their models using Insdye BIOS's.

 

Yup... The Helios has indeed 16MB ROM chip (128Mbit)... however, as we know that is still enough for Zen 2 (and Acer's BIOS UI is quite primitive - its on par with pre-UEFI interface - nothing fancy, just a basic blue/grey/white options in DOS style).
Meaning Acer could have given those of us who use this machine an option to upgrade to Zen 2 (and eliminate Zen 1 out of the microcode entirely - which wouldn't even matter since the machine comes with Zen+ anyway).

Some X570 mobos ALSO come with 16MB ROM chips too (per that google DOC)... and AMD mentioned ALL B550 and X570 mobos will have Zen 3 support (those smaller ROM chips might have to cut out some CPU's out of the equation - such as Zen 1 family - which would be understandable, but still doable - and if that happens, then it just means that Zen 3 support could fit onto 16MB ROM chip which has a basic UI and by cutting out Zen 1 support).

 

Is it really the mobo mfrs (mfr is short for manufacture or manufacturer, whereas mfrs is short for manufacturers, generally) when they didn't need larger ROM chips for Intel? AMD was still seen as budget and didn't have Zen 2 performance to smack Intel with.

As such, why would they assume they need a 32MB chip. Also, amd added new boost features and algos each generation. Implementation takes some space.

So I do not put it on them. AMD can add it to spec. Once it is spec, they have to have it to be in spec. So this falls on AMD, IMO.

Instead, this is closer to the trx40 boards, except users have been warned what board is needed. I applaud that.

At the same time, forcing a potential board upgrade on mainstream consumers during a financial crisis and with being the last gen on DDR4, I call that a bit silly. But it does stick with the same claim they made before, so...

Sent from my SM-G975U1 using Tapatalk

 

I would say that at first (just before Zen 2 release), it was indeed the motherboard manufacturers fault for not giving mobos sufficiently larger room chips (such as 32 MB).

They knew that AMD planned to keep supporting am4 through 2020 because it was already announced years ago.

I think they should have prepared for that... Although on another note, people might consider themselves lucky they got those updates in the first place as it was up to the manufacturer to integrate the microcode into the bios or not.

However, in this latest move, it was AMD that said they won't support anything less than B550 and X570 (despite both coming with 16MB and 32MB ROM chips - same as previous mobos).

So AMD excuse on ROM chip size being insufficient is not valid (As you said, especially during the coronavirus pandemic which is causing so much financial issues and Zen 3 being the last iteration for AM4). To me, this seems like a bad move (still, unlike Intel which seemingly changes motherboard with each CPU release, consumer AMD chips [except Threadripper] were supported all the way from from Zen 1 to Zen 2 from the first B350/X370 - with some exceptions).

 

Now here is a video that is important in understanding 6 and 8 core performance of Zen 3 aside from the IPC gains.


AMD would not say 12-15%IPC gains meaning only on latency sensitive apps. So, the advantage the 3300x has will be on Zen 3 along with that IPC gain, hopefully, which would be a monster jump in latency sensitive workloads.

Sent from my SM-G975U1 using Tapatalk

 

This was my understanding as well.
IPC gains do NOT include latency reductions/improvements - which means that latency sensitive software would benefit from both latency improvements AND IPC gains (and potential small clock bump ups).

EDIT: This is very accurate for Zen 2 mobile parts... because the 4700U and 4800U actually surpass Ryzen 2700x in raw performance seeing how Zen 2 mobile parts are essentially monolithic designs which eliminate latencies across cores.

 

Sometimes latency impact is counted, sometimes it is not. What matters especially is productivity and corporate uses. If you notice, many rendering uses were not as latency sensitive compared to gaming. Zen based chips, regardless of generation, are made for servers. Period.

With Zen+ to Zen 2, latency WAS part of IPC. How? They changed the design to use the I/O chip. By doing so, and standardizing all CCX to CCX through the I/O die to better be utilized by the scheduler, they incorporated that change into IPC because it belong with scheduling effected nearly all tasks.

Here, the change did not effect changes in scheduler activity, etc. Because of that, the reduced latency is very workload specific. If your workload is not latency intensive, like keeping the data in cache like tile based rendering, seen in the cinebench and blender workloads, you see little difference. In compiling like the clang chrome test, we saw a lot of latency sensitive behavior.

Because of this, if AMD claimed 15% IPC, but it was only in latency sensitive workloads, the corporate users could have issues regarding recently built goodwill. Granted, they always do testbeds. But, it still would see benefits for only some corporate users and would give little incentive to go to Zen 3 epyc.

Also, GN saw 17-35% performance increases in some latency sensitive workloads, exceeding the 12-15% range previously given. So my thought is the IPC is workload agnostic, but the latency specific tasks are greatly increased. Otherwise, they are averaging latency sensitive tasks with non- latency sensitive tasks to cut the 17-35% down to 15% (also a real possibility).

This also cuts the latency on multichip cpus as they do not have to jump off chip to another CCX (since 8 core CCX per chip with 8 cores per chip), which can improve performance.

So this is a possibility, not a guarantee, that the benefits are latency agnostic. Time will tell.

Sent from my SM-G975U1 using Tapatalk

 

I am sure if they wanted there could be a bios made for the emergency recovery that does not require a CPUto install. With this bios eliminate all older CPU compliance and just have the new code, you no longer would need the compliance with the new CPU installed.

 

The Canadians will get their first hand on the Ryzen heart Legion5, compare to other markets around the globe...
https://www.lenovo.com/ca/en/laptop...Legion-5-15ARH05/p/82B5CTO1WWENCA0/customize?
...however, they'll hv to pay a little more to be first in queue...

 

When I built my desktop I originally had an ATX x570 board picked out. Then I decided to go matx with a b450m and switched everything out

Went with a 1600 AF thinking I would probably just upgrade to Zen 3 when it rolls around. I was foolish

My PC still performs very well for what it is, but I think I'll probably list it for sale and not be too upset if it doesn't sell.

 

I'm just loving all the YouTubers laying out on about this.

I want to address a single argument which makes no sense here but did with TRX40: that AMD NEEDS to in order to add new features. That doesn't apply to AM4. Why? X570 has been out for a year. B550 adds nothing new and isn't the X570 high end. So you are getting NO ADDED BENEFIT of "new features" with the decision to limit which boards these chips are on.

With TRX40, the VRM was part. They argued ram, which was a lie because they figured out how to do it on AM4. But they had PCIe 4.0 with all those lanes, etc. Here, what do you get with the restriction? NOTHING! Not a darn thing.

I cannot point to a single new feature you get with this decision because the boards have been out forever and no change occurred.

So, it is fine to accept it, but do not believe bad arguments!!!

Sent from my SM-G975U1 using Tapatalk

 

The issue is that AMD asked and sell although there by the chips. Also the OEMs need to be able to sell the motherboards they market. This is going to be the last of the AM4.

 

Sent from my SM-G975U1 using Tapatalk

 

As you know, despite what that AMD graph says, both B450 and X370 mobos (at least plenty of them) got a BIOS update for Zen 2.
So, regardless of AMD claim, it may be possible that OEM's will still integrate support for Zen 3 into their mobos (could be they will erase support for some CPU's such as Zen 1 and Athlons, or downgrade UEFI graphics to go about it).

EDIT: Scratch that. Looks like AMD treated that as an optional update for previous mobos. This time they don't even plan on doing that.

 

It will be interesting to see what the next generation chipset looks like, but the 3950X in particular is running into memory bandwidth issues on dual channel memory. So certainly on the high end there's a case for a more memory-capable (and possibly I/O-capable) chipset than the X570. Of course, that doesn't really much apply for the lower-end chips.

 

That DOES NOT MATTER on Zen 3! Mainstream is not getting a new socket this year. DDR5 is slated for Zen 4. So although I agree it is memory limited and they need to do something about it, along with questions on if DDR5 will be enough extra bandwidth (i think mainstream should have single slot channels with quad- channel memory or 2 channels of 8GB HBM2E supporting on chip with dual channel DDR5 moving forward, but...), but either way, it is moot.

This generation is the last on AM4. AMD doing this now to mainstream is a mistake. They will self- own on this while Intel is down, limiting their success by the time the empire strikes back, so to speak.



Sent from my SM-G975U1 using Tapatalk

 

Thr Ryzen 4000's journey into gaming body doesn't bottleneck @ just Legion5!
This Reviewer/Retailer based in China gives u a comparison review and sneak peek into the internals of the Ryzen 4600H and Ryzen 4800H vs Intel 10th Gen i7-750 and i5-300 that will come in the Legion 7000series as R7000 and Y7000!

4600H vs i5-10300H with same GPU of GTX1650


4800H with GTX1650 vs i7-10750H with GTX1650ti


Both comparisons yields marginal benchmark, especially with the surprising result on GTX1650 vs GTX1650ti.
Hence, the budget Ryzen based Legion 7000 seems to be a good buy worth looking out for... probably in late July for international release!
Chinese subtitle is included for viewers convenience, that doesn't limits to Chinese reading or speaking folks. Click YouTube's subtitle settings and go Auto-Translate to the language of ur preference.
(note that YouTube's auto-translation may produce the type grammars to different language that may fail to meet viewers' expectation.)​

 

Dell G5 15 is now available in Intel 10th gen Comet Lake-H and AMD Ryzen 4000 Renoir flavors, but AMD as usual gets lesser oomph

https://www.notebookcheck.net/Dell-...-AMD-as-usual-gets-lesser-oomph.465527.0.html

"Now, take a look at the AMD version and you will find that the laptop is gimped in several aspects. The Dell G5 15 5505 SE, first announced during CES 2020, offers a choice between the AMD Ryzen 5 4600H, Ryzen 7 4800H, and the Ryzen 9 4900H. All SKUs offer the AMD Radeon RX 5600M GPU as standard making this probably the first Ryzen 4000 + Navi 10 combination. MSI had the Alpha 15 with a Radeon RX 5500M but that was paired with the Ryzen 7 3750H. Both the Intel and AMD SKUs of the Dell G5 15 will be available from May 21 at prices starting from US$829.99.

While the basic specs of the G5 15 SE are just about right for FHD 60 fps gaming in most modern AAA titles, this model does stand to lose out on the 300 Hz display, some of the dual-storage options, chassis lighting, and Killer Ethernet.

Sure, the 300 Hz display is probably an overkill for the RX 5600M and the chassis lighting is more of an aesthetic than anything performance impeding but here's the thing. OEMs don't yet seem to be willing to outfit AMD models will all the pizzazz that their Intel variants stand to get. In all likelihood, OEMs are testing waters to gauge the sales of AMD laptops before giving them the same spotlight as the corresponding Intel SKUs."


'OEM's testing the waters' is a bit of an old excuse here that simply doesn't hold ground (because they've been 'testing the waters' like this for a LONG time now).

If an OEM doesn't offer an all AMD system with as many customization options like with Intel, then customers might be less inclined to go for the all AMD system BECAUSE of lack of options.

It's like Dell intentionally wants to sabotage an AMD system right from the start (and this wouldn't be the first time... it dates back to Zen 1 and + too).

However, looking at the actual Technical Specifications for the AMD version:



It doesn't look THAT gimped.

The 5600M looks like it should be able to match/beat RTX 2060 and RTX 2070 MaxQ (although I DO think DELL should have included an option for 5700M).
Realistically the RTX 2070 MaxQ is on par with RTX 2060... except is requires less power?

The display CAN apparently be upgraded to 144HZ version on AMD system (whether the 144HZ option has a better quality display overall and comes with FreeSync, I have no idea... perhaps, but hopefully, the review will cover this and not just the base model - and also, perhaps DELL will make this info available).

Overall, I wouldn't be too upset about lack of 300HZ display (especially when most games on this GPU or even RTX 2070/RX 5700M won't even get 300 FPS to begin with under 1080p - unless they are much older titles).

If the 144HZ option has high/decent color range, great brightness and other overall specs, then it shouldn't really be a problem (nothing prevents OEMS from making high quality 60Hz and 144Hz displays with overall great brightness, color range and contrast - but I do agree that 144Hz display should technically be a minimum for these GPU's and 300Hz displays at least on offer if they are available for Intel/NV).

EDIT: Gigabit Ethernet option for NV, but I'm not sure people will be bothered with that as external GPU's for laptops aren't something most would need seeing how the 5600M should be powerful enough for the next few years (4 or 5 perhaps before people decide to upgrade?)... heck my Vega 56 would be equally powerful and I can OC it (without exceeding its 120W limitation) to match RTX 2070 (desktop) while keeping in mind that Vega 56 still exceeds both 5600M and RTX 2070 (desktop) in compute performance by quite a bit.

Dual storage does seem a bit gimped.
No apparent option to pair an SSD with a HDD - so, users would be limited (does this mean the laptop lacks a SATA HDD bay completely or just that DELL isn't offering a HDD for storage)?

If there is an (empty) SATA port inside the laptop for a HDD (like a SATA HDD bay), then its not so bad. You can get your own and install it inside as a second or third drive for extra storage (although these days it seems just a tad cheaper or at least same price-wise to have the OEM integrate the HDD or other hw inside for you - it used to be more expensive).

Its a setback yes, but unless the actual HDD bay with a SATA connection is missing, then I don't think its a [large] deal breaker (Dell would only have to let people know they can install their own HDD by opening the back panel without breaking the warranty - which certainly seems to be the case when they mention the laptop can be user upgraded to 32GB RAM - although why they aren't offering this as an option from the start is a bit of a question - plus, RAM should have specs in it such as speed and timings - also, any OEM should be able to offer high frequency and low timings RAM for laptops [which can be a bit more expensive yes, but honestly, its not a huge expense]).

If DELL offers Intel Wi-Fi 6 AX200 as an option then I personally wouldn't be bothered by lack of Killer Wifi (performance-wise and board-wise they would be virtually identical if both are of the latest model... the only difference is in branding and the fact Intel is cheaper - no real loss there).

EDIT: I have now noticed that the Wifi DOES come with Killer option 1650x 2x2 AX model (which is the only option on AMD system, whereas on Intel/NV version of the laptop you can pick the original/cheaper Intel Wifi 6 module... this probably increases the price of AMD system, but not by much I hope (in fact, the difference in price shouldn't be huge as they are basically identical cards with pretty much same performance/capabilities and only different branding and some alterations in 'features' on the Killer model - certainly not something worth the price premium Killer wants).

Not too bothered by lack of lighting options... so long as the keyboard has back-lighting (obviously)... everything else is more or less eye-candy that's not really needed (light up the relevant bits, leave everything else out of it - no need to go overboard).

EDIT: Technical specs show the keyboard IS indeed backlit and has same options like Intel/NV... again, not bothered by lack of other lighting options for GPU's as I don't see them as relevant (and I doubt many people would).

Battery could be better (such as 73WHr and 91WHr options, but at least DELL offers a more powerful option of 68WHr compared to the MSI Bravo 17 which is limited to 51WHr).

Barring any inferiority in hardware quality that could be revealed in the actual review of the unit, it seems that DELL hadn't exactly gimped this unit as much as they did Zen 1 and + options)... also the price seems pretty good actually (especially when you factor in that the cheapest AMD option is on-par performance-wise with the more expensive Intel/NV option in the same chassis - and is likely to EXCEED the Intel/NV with higher minimums and possibly higher overall average FPS due to more powerful CPU option?).

The only thing left is to wait for a review and see how the cooling (and the rest of the system) stacks up (hopefully, DELL has done AMD justice on this one - or if not, then the reviewers should try a repaste with something like GC Gelid Extreme or Kryonaut to see if that improves that situation - I know that DELL had issues with thermal applications from factory in the past).

 

Really looking forward to this one and its price point

 

According to the (trustworthy?) leaked benchmarks the combination of a RX5600M and a 4800H scores better than a RTX2060 and 10750H in separate Firestrike graphics and physics tests, but it scores worse in the combined test. This could point to thermal or power issues when CPU and GPU are both under heavy load. In any case, quite promising!

 

From the link...


Last years MSI GP65 Leopard 9SE Laptop. 17% higher graphics score. The RX5600M isn't even close.

https://www.notebookcheck.net/MSI-G...een-on-a-mid-tier-gaming-laptop.431863.0.html

 

Looking like AMD's cpu is fine, but still need those Nvidia gpus paired to make them worth it... lol

Sent from my SM-G975U1 using Tapatalk

 

A pure Red Blooded Breed? GOOD!!!
KUDOS TO DELL!!! However to compete with one another pure RED BLOODED BREED, DELL gotta consider a bigger 17incher with bigger body for better air ventilation... to compete with MSI Bravo17!

 

All red is not good enough yet. I haven't comparatively priced, but if the same price as with a 2060, it just isn't good enough!

AMD did this same crap with Zen and zen+, forcing the pairing with Polaris and then Vega. Forced pairing isn't letting the market work, its saying if you need the CPU power, you have to accept an inferior GPU. That will reduce sales and adoption rates.

Also, for perspective, Dell wanted nothing to do with AMD. Why? That Intel money. This is why for epyc builds, they originally planned only 3 for Zen 2. The exec made derisive comments about AMD. Months later, he was forced to backtrack and say they were offering 9 AMD builds out of 50 some builds the offered.

Dell also was implicated in the pay to not use AMD CPUs over a decade ago when AMD had the better CPU (before its dark years).

So...

Sent from my SM-G975U1 using Tapatalk

 

Just to make something clear: you don't need a 17" chassis to get decent/low temperatures (for that matter, a 17" chassis won't really matter if the cooling is badly designed)... HOWEVER, I AM partial to 17" for the screen size (along with possibly better thermal control and bigger battery).

So, yes, I would agree that Dell should consider a 17" chassis.

Who says the 5600M is an inferior GPU? It has a rather low TDP (and could be potentially undervolted for even lower power draw) - its overall specs are IDENTICAL to the desktop version of the chip sans the slightly reduced clock and VRAM frequencies.

The Intel/NV option from Dell offers either RTX 2060 or RTX 2070 Max-q (both of which perform pretty much the same).

Given 5600M specs, it will probably exceed both the RTX 2060 and 2070 Max-q thanks to the Zen 2 APU options.
Also, please don't use 'lack of RTX' for 'AMD GPU's are inferior' because I'm tired of that argument.

Also, how exactly did AMD force anything here?
It was DELL's decision to integrate both Zen 2 and mobile Navi.
Just as it was Acer's decision to use Zen + and Vega 56 in a desktop replacement (which turned out to be one of the most powerful desktop replacements on the market with best cooling [even today] and V56 in that unit is a monster of a GPU which can easily be OC-ed and undervolted without going over its 120W TDP to match GTX 1080 and EXCEED it radically in compute performance).

AMD an Intel can certainly influence certain design choices, but we know that neither AMD or Intel have much of a choice in saying to OEM's 'pair it with best possible screen, GPU, etc'.... that much is up to the OEM (and yes, I'm aware Intel's money probably makes conditions more favorable for Intel/NV solutions).

You want a Zen 2 and NV option? They exist... in fact they pretty much DOMINATE this release cycle in laptops from other OEM's
Just because Dell hadn't released something that's similar (yet) doesn't prove anything.

And freaking heck, we NEED more all AMD options on the market. I'm SICK of seeing laptops with only NV dGPU's as an option or low end (and outdated) AMD dGPU's... give me mid/upper-mid range options like 5600M and 5700M for crying out loud paired with decent/powerful and QUALITY hardware.

And yes, we are aware of Dell's history and how it gimped previous AMD products... we still don't have the G5 15 SE 5505 on hand, so we really can't say anything yet (conclusively) about its screen or cooling quality/performance.
If Dell messes that up, then at least we should know that from the reviews and understand that Dell's 'AMD experiment' was designed (by DELL) to fail from the start.

 

Specs don't mean anything in paper if the performance isn't there. Papusan have some evidence that the Intel with the 2060 outperformed the 5600m paired with the AMD CPU. The AMD CPU performed real well on its own. That combined with majority of the designs being with Nvidia cards say that is the better pairing.

That is data driven. I do not care if it matches the desktop performance. I even have an out on the basis of price, as if there is enough price savings, then the laptop is worth it. But you didn't seize on that line of inquiry, which would resolve any conflict immediately.

As to not forcing the pair, are you familiar with offering discounts to move product. AMD needs OEM adoption, and wants not just CPU, but GPU adoption. So since they have a good CPU, they can cut a discount coming from the CPU margins to move their GPUs for market share. Is it exactly forced? No. But for Zen and zen+, it may have been the discount level that got those laptops built in the first place. In that sense, it could be considered a forced pairing.

For Zen 2, it is a beast. AMD is taking mobile market share. It's so good, along with Intel's manufacturing and supply woes being as they are, so that the AMD CPUs would be used no matter what, even without the extra discount. To make those more attractive, especially getting over the perceptions of consumers that AMD is budget or unreliable or..., they then pair the CPU to Nvidia GPUs. That gives a higher status.

Moreover, this isn't about RT. This is about real work performance. That means best frame rates or ability to do the task assigned (AMD drivers still have issues, open CL/GL not as developed as Cuda, etc. ) takes precedent. This may change with RDNA2 and the 3000 series of Nvidia GPUs. We'll find out.

We all assume the zen 3 is going to be good. No clue if tiger really will be or not. So that fight is worth watching also for mobile (desktop, I have less faith in Intel until 7nm).



Sent from my SM-G975U1 using Tapatalk

 

I'd very much like to see the performance analysis Papusan has about Intel/NV combo being 'faster' than Zen2/Navi in this DELL laptop (especially because I hadn't seen anything like that as of yet since DELL hadn't released anything officially [at least not to my knowledge] - and besides, early tests don't mean too much without a finished product being tested at full capacity [we know earlier samples usually run at lower frequencies or are still being in testing phase] ).

OpenCL/GL not being used (or being neglected) is not AMD's fault though... its the developers choosing not to use it even though its just as capable as CUDA (and we know the reason behind this is because NV has deep pockets and has paid devs before to use CUDA instead of OpenCL - that's the main reason).

As for the drivers... what issues specifically?
For work (and gaming) you can more or less easily use AMD Radeon Pro Enterprise drivers which are pretty stable/reliable (and besides, if you're using productivity software, the Enterprise pro drivers would work better for that software)... also, AMD took care of the drivers mostly on the consumer end as well.

Correct me if I'm wrong, but AMD integrated mobile drivers release cycle into their desktop driver release cycle back in 2018... so new releases are far more frequent and I hadn't seen recent reports of mobile drivers from AMD having problems.

As for desktop driver issues... those were largely ironed out, and to be perfectly blunt, NV had/has its share of driver problems (which aren't as largely publicized). All you had to do is follow same search standards for problems which manifested on some AMD GPU's and apply them to NV to discover their forums are teeming with similar problems.

Incidentally, a much larger % of the people who bought AMD hw did NOT experience problems. Yes, AMD acknowledged the issue and set out to work on it... so what?
Isn't that what companies who produce GPU's are supposed to do and that some problems crop up when you're running a robust ecosystem like a PC which has multitude of different hw?

Incidentally, since I've been using all AMD desktop replacements, drivers hadn't presented problems for me (except for the consumer drivers not playing well with my Vega 56 as coming out of sleep mode producing static on the display not being fixed - however, I've been using latest Enterprise Pro drivers which don't have ANY issues and have been running stable for me (and I hadn't lost any performance in games - plus, these Pro drivers are far more useful for my productivity software/uses).

Also, any business which decides to use Intel/NV or all AMD laptops (or AMD/NV, or Intel/AMD) would probably opt for professional drivers anyway because they know they are more reliable than consumer grade drivers (regardless which company you choose), or would basically stick with OEM provided drivers and update them only if/when they find it necessary).

 

AMD, on the GPU side, is infamous for poor drivers where the specs are better but performance is worse or does not match the numbers.

 

For example, their DX11 driver overhead, which has been an issue going on almost 7 years now and will never be fixed.

Radeon driver-limited to 148 FPS here:


Meanwhile AMD fans be like but muh DX12/Vulkan!

 

Taking reference on the thickness of the current G5 5590

@ 23.7mm thickness/height
vs
21.6mm thickness/height on


I wouldn't be surprise that "just 2mm shaved off" will be the comments from those who seek thin laptops,
but... with such internal view on the current G5 15(5590)

...I couldn't imagine what kind of heat will build up internally in that new 5505 tighter space...
Hence the thought of "bigger interior, fatter height,=better ventilation" comes to mind.​

* Whatever spins money, DELL, as well as other makers build it.
My first laptop is a DELL Inspiron 1501 with Turion TL-50. Back then there were other AMD CPU flavors to choose on. Celeron's direct competition from Sempron and Athlon. And mine was a direct loggerhead with Pentium then. It was until when Intel enters into Core2Duo stage, DELL DUMPED AMD!

** I've seen more AMD+Nvidia combo in the market than AMD+ATi Radeon combo.
The most recent ALL RED combos i've seen goes to the Ryzen 3##0H with RX5#0, that found in Acer Nitro 5 and Asus TUF Gaming FX models. But most other majority in the market still pairs Nvidia with the Ryzens.

*** Considering that the dGPU comes with an extra "0" and performance on par with Nvidia's GTX 16#0ti or max-q series...
https://www.notebookcheck.net/AMD-Radeon-RX-5600M-GPU.449471.0.html

With some improvements expected on the RX5600m and starting price on par with INTEL+GTX1650 configuration,,,
https://videocardz.com/newz/dell-g5...n-4900h-and-radeon-rx-5600m-launches-may-21st
I would consider the price is competitive...

Ahhh... That explains what i saw here...

...when the reviewer laptop is on battery while executing the CineBench20, the wattage drops to 15w "TDP", when plugged on and executing the same benchmarking program, the "TDP" shoots up to 45w?​

 

This is pretty normal for battery conservation. Not being a gamer this is perfectly acceptable behavior.

 

TDP doesn't actually change, because it is just a rating that is only valid under the same testing conditions that intel/amd used when they defined it.

A TDP of 45w remains at 45W regardless of the way that the PL1 changes, because the TDP should be relevant only for the cooling system designers. I used "should be relevant for" because even the cooling systems manufacturers can't make anything useful just with the TDP rating, without having the rest of the specs and guidelines.

A CPU that is rated at 45W TDP remains a 45W TDP CPU, but the cooling system should be designed to be able to dissipate a minimum of 45W of heat under the same conditions used when the TDP was defined (even if the cooling sys manufacturers use a different formula to calculate the heat dissipation potential).

Raising or lowering the PL1 doesn't affect the TDP rating, because you are not testing it under the conditions in which the TDP is defined, so it is incorrect to substitute the PL1 with the TDP. A better way to say it: testing a 45W TDP CPU at a different PL1 limit (raising the PL1 limit can be indirectly considered overclocking).

Here is a better (more detailed) explanation of what TDP represents:

 

Thank you for arguing in kind. I do appreciate it. But I will have to point out, even though it may have only applied to desktops and servers (nothing I've read clarifies, just says overall computers), this article here regarding Intel's payments to Dell.

https://fortune.com/2007/02/15/suit-intel-paid-dell-up-to-1-billion-a-year-not-to-use-amd-chips/

Sent from my SM-G975U1 using Tapatalk

 

Thank u for providing that link proving that Intel playing dirty is not a "myth" but an open truth!
However, looking back, without such dirty trick employed(where i believe doesn't just applies on server CPU and desktop CPU), there wouldn't be failures of the Bulldozer and Excavator, that drives the effort in engineering to create the Ryzen.

 

Eh, bad leadership during Phenom gave bulldozer.

AMD bet on CMT, which no one programed for. SMT won.

For Ryzen, most of that tech was shelved by 2012, later revived by Lisa Su.

I do not like to blame Intel for AMDs failures there. AMD sat on its laurels, and the CEO saddled AMD when purchasing RTI. This impacted cash on hand and research. This was also around the time that GF was spun off (good thing except for agreements to use them over whomever had the best process).

So Intel played dirty, but let's not forget the numerous times AMD was its own worst enemy...

Sent from my SM-G975U1 using Tapatalk

 

I am interested in a 17 inch 4k 4900h laptop, hope to see one soon. Would love to see a rtx2070 or better but not a deal breaker here.

 

From the Wccftech article:
The following are some 3DMark GPU benchmarks that were leaked last week.

• RX 5600 XT 3DMark Time Spy - 6446 Graphics Score, 5884 Overall Score
• RX 5600 XT 3DMark Time Spy Extreme - 2815 Graphics Score, 2576 Overall Score
• RX 5600 XT 3DMark Time Spy Extreme - 2791 Graphics Score, 2891 Overall Score
• RX 5600 XT Fire Strike - 18906 Graphics Score, 14341 Overall Score
• RX 5600 XT Fire Strike Extreme - 9198 Graphics Score, 8087 Overall Score
• RX 5600 XT Fire Strike Ultra - 4738 Graphics Score, 4672 Overall Score
• RX 5600 XT Fire Strike Ultra - 4756 Graphics Score, 4688 Overall Score

It seems that compared to the MSI GP65 as tested on NotebookCheck, the RX 5600M scored higher in graphics than RTX 2060.
Perhaps I'm reading the results wrong, but where are you getting '17% higher score" for RTX 2060?


EDIT: Also, those are 5600M leaks, are they not?
Which means, they are not necessarily indicative of final performance metrics, and we'll need to wait for reviewers to actually review the 4800H/4900H and 5600M in full.

 

The ID of the "Generic GPU" is not clearly identified... so... can that be the RX5500m that's used in the MSI Bravos? Or even better, iGPU of the Ryzen 4800?

One thing for sure... that result, is reaching RTX level...

 

What you post is results from the desktop cards...
https://wccftech.com/amd-radeon-rx-5600-xt-gpu-benchmarks-leak-faster-than-gtx-1660-ti/

El cheapo Walmartbook i7-9750H + 2060

Re-read my post... http://forum.notebookreview.com/thr...laris-navi-gpus.799348/page-764#post-11014039

RX 5600M https://wccftech.com/amd-radeon-rx-5600m-6-gb-gpu-benchmarks-faster-than-rtx-2060/

2060 Mobile (17119 graphics score).