@@BonusCrook Lmao that's a huge success. They save so much money with that tech selling the cards to you with a way higher margin. The way to enjoy rdna3 is to buy amd stock instead of the cards.
I agree with alot of the comments about how the 780m is not a huge improvement over the 680m in averages, which is disappointing, but the more important thing for me is the 0.1% lows, which where up across the board by 5+ fps, which is more important (imo) than increading average FPS up alittle more. Keep in mind the 780m was also memory choked, it supports even faster vram, which would have helped it. Either way, nice to see a comparison.
@@bulletpunch9317 this question makes no sense to me. Maybe rephrase for clarity? My point is APU is a package deal of CPU and iGPU. You cannot take CPU from older one and put newer iGPU on it, so it does not make sense to ponder whether improvements come from CPU or GPU gains.
Looks like power consumption of 680m is capped at 54W and 60W on 780m. I think that if they were both limited to 54W consumption, then the test results would be the same.
@@mparagames they both have 12CU's. optimization is possible and takes place. but still I tend to think that the performance gain is provided for the most part by the increased power limit.
@@mparagames it is possible to find out what caused the increase in FPS only by capping the consumption of both the GPU on 54w or by capping the FPS at a certain value and measuring the average wattage.
@@LLIEHEBMEPVA i mean, in general for these APU the power scaling is so good that 6W at this power level doesn't make that much of a difference, as power and performance aren't really linearly related. 6W (11% more power) for sure wouldn't be enough to boost the clock speed by 16%, the reason this boost is possible is more rather due to lithography improvements
@@LLIEHEBMEPVA yeah it is, but i don't have these devices to compare.However, from the video it is possible to see that it wasn't only the power advantage that was at play here, tho we don't know hoe much it impacted the results exactly.
You know what I love most about you mate? That you don't only promise that you're going to make a video that you say you will, you make it way faster than anybody ever expects. Major props to you. You not only deserve 500k subs, but double or triple that. May the world bestow upon you the most riches you could ever dream of 💪
I think the 780M is getting bottlenecked by RAM speed at this point. Would love to see this same test but with 7200MHz RAM, now that the DDR5 prices have started to come down.
@@MidWitPride you're not getting 7200MHz SODIMMs anytime soon. At best, you could hope for LPDDR5/X soldered directly to the boards with the APU. Obviously, that would require a different PC from the one used in the video.
Someone who's not afraid to do some benchmarks even if the margins are tiny. It puts my mind at ease that I don't have to upgrade my 680M handheld any time soon. AMD have done a wonderful job at bringing gaming to the masses without breaking budget on a discrete gpu.
I would disagree. You could always play contemporary games with AMD iGPU on low settings. Also, these are best, most expensive models, and for that money you could buy low-mid range CPU and GPU that would run games better.
@@aleksazunjic9672 BS... Compare them to Intel's iGPU's. New iGPU's are way are better than old budget' 'flagmans' like GTX 750Ti or 1050Ti. Post author had a iGPU 'evolution' in mind, not a direct comparison between iGPU and a discrete card (it would be pointless).
@@aleksazunjic9672 But in terms of efficiency this tiny systems are very much better. I don't know any classic PC which reach this performance with only 50-60 w.
@@aleksazunjic9672 I agree, compared to PS5 APU, whose graphic performance is at par with a desktop RX6700 10 GB, a 780M iGPU really isn't that powerful. And far too expensive. AMD could deliver an adapted PS5 APU to desktop PC if they wanted to, couldn't they
@@guilhermehenrique-zj5ttthe uplift in the average fps is ok, but the 1% and 0.1% lows are the real progress. Thase uplifts by themselfs are the real mvp since it means more stable and less noticable frame drops.
@@Gun10001 I wonder how much of the uplift in percentile lows is actually from the newer CPU or memory configuration (timings, latency, ranks). I somehow doubt that the GPU is the one boosting frametime consistency.
I just wanted to say thank-you so much for your inspiring work, especially concerning these Minisforum PCs. Long story, short - I somewhat desperatly needed a little PC like this and your videos inspired me to get a UM690 with a Rysen 9. I know that my new PC has been surpassed by this new model but it wasn't released in time. The fact that it can play some games with some compromises is enough for me 🙂. 3 weeks on: - Still ebjoying ny new 690 - Got to tweak Windows 11 a bit 🙂 Thanks again!
Very cool, interesting how well the 780m is doing in such a mini formfactor. Was wondering what the 780m would look like put up with a Ryzen 5600G or 5700G APU. Just to see how far the gap is between mobile chips and desktop APU's in 2023... Anyway hope you Enjoy.
And now that RAM prices are coming down too, i'd like to see some faster than 5600MHz RAM on these things tested out. You can get 32GB of 7200MHz DDR5 RAM for less than $200 these days.
@@MidWitPridethe original poster very clearly states he wants to see the 780m up against the 5600g or 5700g. Specifically to see the difference between desktop and mobile chips, now, in 2023. No mention of any future CPU/APU at all.
Hmm, not bad, now I really want to see how these fare against older discrete GPUs like the 1050, 1050Ti and (2GB) 960. The latter is what I have in an old tower build, and yes, it is the 2GB variant.
It will obliterate them. My 680m on 7735HS is on par with my old R9 290x. You can also adjust VRAM in the settings, as vram is basically ram and DDR5 is very fast indeed (mine is 4800mhz). Impressive APUs.
Thank you for this. 10% is very much inline to what our knowledge of RNDA 3 improvement to be when looking at their desktop gpus. those 1% lows are incredible though, many games are much more playable.
Finally a decent comparison between 680m and 780m. I'm thinking of buying the minisforum EM680 or wait for the EM780, but the price difference will be approx $200. Do you think 16 gb is enough or should I get the 32 gb for $50 more?
igpu will take a chunk of RAM as VRAM, so instead of 16gb system ram you'll have smth like 12gb ram + 4gb VRAM. I'd advice going with 32gb ram, if possible
get the 32GB remember you can easilly use 16GB for games alone... you need "VRAM" too... so get the 32GB and mark like 8GB for the APU and you are set...
These results are impressive to me from the point of any integrated graphics, but disappointing for the theorized uplift inperformance. I think retesting by increasing the usable vRAM limit from 4 GB to 8 GB a must. And a possible retest by upgrading the RAM kit in both 2 32 gigs and then having at least eight but up to 16 gigs of the ram in the test. I think that would be a good way to see the differences. and of course Do a comparison with the winner of these two to a 13th GEN Intel integrated graphics of any model
Hmm, the 780M doesn't add as much performance compared to the 680M as I expected considering the sizable 400MHz clock boost. It does seem a bit more stable though, especially in GTA 5, the 680M's clock speed was all over the place, but the 780M held steady.
RDNA3's jump in performance was overall disappointing compared to RDNA2's in discrete graphics cards and these APU results reflect that as well. The difference could easily be attributed to increased CPU grunt, power consumption and clock speed favoring the 780M. This whole gen just screams "Skip Gen" to me, between the outrageous prices and disappointing performance bump
RDNA3 was supposed to be where they used chiplets to get more shaders on the card but they decided to only do 1 GCD for some reason. Really strange decision to throw away their advantage.
@@BonusCrook there are enough problems with RDNA3 already. 100w+ idle power consumption with multiple monitors, some of the VR games being broken, AV1 decode/encode sometimes not functioning correctly.. it's clear that the Radeon Drivers team didn't have enough time/resources to roll out proper support for RDNA3 even with single GCDs, 2 GCD designs would be a mess
Considering both can play triple A titles at 1080p low/medium settings is awesome.. If someone said integrated gpus would be capable of 1080p gaming a few years ago people would think its a dream, but now its very possible.. If the person doesn't care about playing at 900p or even 720p then its definitely possible to play at 60fps high.. Sadly nowadays 1080p is considered minimum, but having fun is the most important, even if you have to sacrifice graphics and resolution to have playable fps.
So i have the beelink ser 6 pro 7735HS with 680M. I also got it only for 350 BUCK! But i was wondering if i should upgrade to minisforum with 780M graphic. What do you think should i upgrade or am i good to go for now.
Would love to see you test them in the future when faster ram is available, because i really think 5600mhz ddr5 ram is kind of a bottleneck(even the 680m will be improved)
Came to say this. The way I understand, these iGPUs really benefit for faster RAMs, so there is probably a few extra %'s of performance to be had there...
@@rhobson it honestly could be more than a few %'s. The 6800u could already perform better with 6400mhz lpddr5 ram and it was a decent uplift. The 780m isnt the double the performance for sure but with 7400mhz lpddr5x ram it will without a doubt perform much better than shown in the video. And also increasing the access to 8gb would help a lot in vram intensive games,which almost all modern games are, but with 16gb of ram it not a good idea(32gb would easily allow it).
I like those small Minisforum machines and I consider buying one for work mainly to lower power consumption. My main PC needs 65W in idle with one monitor and 95W for two monitors. It's just a waste of energy and in summer makes my room even hotter.
It* remarkable that the these APUs (particularly the 7940hs w/780m) are in Laptops now as well, Gaming laptops at that, where they can now allow these behemoths that once can only last for about 3 to 4 hours at most even with just playing indie stuff or web browsing, to last a whopping ten hours without even touching discrete graphics at all!
Just FYI laptops have used some form of dual graphics way before this CPU came by man, I had GTX cards along with Intel HD some 15 years ago....given it wouldn't stay on for as long as today's laptop can.
According to his intro screen, he was running both at 5600MT, but I would have liked to see that in the overlay. I have only been able to convince my 690S to run the memory at 5200MT (default is 4800MT), while the 790Pro defaults to 5600MT (providing you have memory that supports that speed).
I have a 6850U with the 680M in a 14" thinkpad and I'd predict in more realistic use cases in laptops there would be very little difference, under 28W STAPM I rarely hit the maximum 2200MHz GPU clock.
The 780M by design and characteristics is at best 10% faster than the 680M. In the new APUs, the cpu is a bigger upgrade imho and I believe that is likely what you are seeing on the increased 1% lows. Quick way to test this is to OC the 6800H to try and match 7940Hs clocks. Or just match the clocks if that is not possible. It does raise a question about price to performance as you can get a 6800H system for about 350 pounds whereas you are looking at at least 500 for a 7840HS and more for a 7940 system.
@@missrochelle1126the new ones yeah. I mean, 500 quid for a whole computer is not that much especially at this level of performance I mean, you can play modern AAA games with upscallers at 1080p looking gorgeous. They pack a lot of performance.
Minisforum has the ryzen 7 7735HS and 680m UM773 for $539 and that’s a steal for a very compact pc with a gaming performance near to dedicated gpus from the yonder years lol
Is the Radeon 680M way better than Nvidia T500 for ~lighter games like DayZ ? I purchased a Thinkpad laptop with T500 and I can still cancel it to get another Thinkpad with Radeon 680M ($400 more expensive).
what i do is low to medium preset then setting the texture quality to high or ultra it looks way better then low textures with less fps drop you can test in a video with few of your cards if you want
What happens if you install 32 GB of ram and give the GPU something like 8 GB of VRAM? Just thinking if the VRAM held them back and you would see bigger differences
Either it's the TPD throttle on the 780M _(on that board),_ or just it's not worth it. Because, taking in mind the new generation _(mind the 2800MHz vs 2400MHZ!)_ - the fps difference just doesn't add up to an expected value, especially considering how triumphant were the "new king" claims for the 7xxM presented by AMD.
Funily enough i own both these processors if you count the R1 Extreme. But since the laptop that has the 6800h in it has a 3050 ti the most I can tell you is the 680m isn't enough to render 2400p HDR content to the oled monitor its attached to. Considering I already have to manually limit the TDP of the 6800 so it doesn't burn itself alive at 100 degrees, I cant imagine adding to that heatload with the igpu.
It's funny because the Ryzen 7600's iGPU in the IO die overclocks to 2800 Mhz @ 1.3V before hitting a power wall, too. Of course that's only 2 CU but that just makes the overclock sweeter since it has a direct and tangible impact on frame rates.
As many have pointed out a comparison with same power limits/frequency would most likely show the results to be equal. Which does bring into question whether the 780M is a "true" RDNA3 a.k.a. generation change. Regardless it is very uplifting to see that both minipcs as well as "budget gaming" can utilize these dGPUs for 1080p even with new games.
the 780m draws a few watts extra because the cpu cores clock slightly higher, the gpu TDP is pretty much the same and for that same power it scales to 2800mhz instead of 2400mhz which is mostly where the performance comes from. But scaling a 680m to 2800mhz wouldnt get the same results as RDNA3 has a different shader layout (yeah both have 12CU but that doesnt tell you anything), RDNA2 has 2x SIMD32 units per WGP but RDNA3 has 4x SIMD32 units per WGP to more efficiently distribute work within the core as RDNA3 of the same shader count has half the WGP's, this was done largely to scale to bigger cards like 7900xtx but it still scales.
Noob question If i play a game capped at 60(not an FPS), will it make any difference vs if i uncap the fps? What's the purpose of a game going over 60?
Precision is one of the more useful advantages of a higher framerate. If it takes 1 second to rotate 360 degrees at 60 frames per second then each step of the rotation is 6 degrees. If you're getting 120 frames per second then each step is only 3 degrees. It doesnt take a second to rotate through a full 360 degrees for most games of course, and the numbers I used in the example are picked at random, but the theory behind it still stands. It also applies still when using a lower refresh screen, even when the framerate exceeds the screen refresh, so there are advantages to the framerate exceeding the screen refresh rate.
Difference not that big which is kinda sad, but on other hand older models gonna cost less now and you can grab something quite fancy during Black Friday etc.
I had considered buying both of these at one point in time. But after seeing the performance I have to say I'm very disappointed. I expected GTX 1650 performance like the products had claimed. Instead we got GT 1630 level of performance.
Wonder how much of a difference the 7940hs vs the 6900hx is making here. The cpu is also much more powerful here. So that isn’t even being factored in the difference between the 680m and 780m. The 780m might literally be a less than a. 5% upgrade
You're using overclocked RAM for the 680M, the native support is 4800MTs. Only 780/Ryzen 7000 Mobile's is 5600MTs. But once more, a much better setup would have been LPDDR5-6400 for the 680M and LPDDR5X-7500 for the 780M, but I know these boxes come in with non-soldereded/LPDDR RAM.
LPDDR5 is slower than equivalent DDR5. Don't just look at frequency it's not equivalent. Memory is 2 things, frequency and timings. If you want to compare, LPDDR5 @ 6400 is between DDR5 5200-5600
It's 800 Euro with 32GB and a 1TB SSD. I don't get the appeal of it. It's laptop parts, but not really portable and costs a lot. A small and efficient desktop PC with a more powerful discrete GPU would cost roughly the same and have better gaming performance.
*We must all remember* *If a desktop or laptop iGPU outperforms their discrete GPUs, then they lose sales on those cards (as they'd be obsolete). This small leap was done on purpose. The 780m already performs as well as a GTX 1660* 😎
Both are relatively capable. The elephant in the room is how little RDNA3 has progressed from RDNA2.
IIRC this is because RDNA 3 is really just an iteration of RDNA 2 and is basically RDNA 2.5.
RDNA3 is AMDs biggest failure in a while. They even made a chiplet architecture without multiple graphics dies. What a joke haha.
I also agree. RDNA 3 left us promising but failed to deliver.
Yeah IT clocks mich higher and has a lot more shader, still it is only a few % faster
@@BonusCrook Lmao that's a huge success. They save so much money with that tech selling the cards to you with a way higher margin. The way to enjoy rdna3 is to buy amd stock instead of the cards.
I agree with alot of the comments about how the 780m is not a huge improvement over the 680m in averages, which is disappointing, but the more important thing for me is the 0.1% lows, which where up across the board by 5+ fps, which is more important (imo) than increading average FPS up alittle more.
Keep in mind the 780m was also memory choked, it supports even faster vram, which would have helped it.
Either way, nice to see a comparison.
So the higher lows are because of the gpu? Not cpu?
@@bulletpunch9317 those are inseparable in the package so not like you can discriminate. APU as a whole is clearly better.
@@klocugh12 do you know which one or not?
@@bulletpunch9317 this question makes no sense to me. Maybe rephrase for clarity?
My point is APU is a package deal of CPU and iGPU. You cannot take CPU from older one and put newer iGPU on it, so it does not make sense to ponder whether improvements come from CPU or GPU gains.
@@klocugh12 do you know if its the cpu that makes the difference or is it the gpu?
Looks like power consumption of 680m is capped at 54W and 60W on 780m. I think that if they were both limited to 54W consumption, then the test results would be the same.
no not really, the 780M should be a bit more power efficient, thus reaching higher clocks still.
@@mparagames they both have 12CU's. optimization is possible and takes place. but still I tend to think that the performance gain is provided for the most part by the increased power limit.
@@mparagames it is possible to find out what caused the increase in FPS only by capping the consumption of both the GPU on 54w or by capping the FPS at a certain value and measuring the average wattage.
@@LLIEHEBMEPVA i mean, in general for these APU the power scaling is so good that 6W at this power level doesn't make that much of a difference, as power and performance aren't really linearly related. 6W (11% more power) for sure wouldn't be enough to boost the clock speed by 16%, the reason this boost is possible is more rather due to lithography improvements
@@LLIEHEBMEPVA yeah it is, but i don't have these devices to compare.However, from the video it is possible to see that it wasn't only the power advantage that was at play here, tho we don't know hoe much it impacted the results exactly.
You know what I love most about you mate? That you don't only promise that you're going to make a video that you say you will, you make it way faster than anybody ever expects. Major props to you. You not only deserve 500k subs, but double or triple that. May the world bestow upon you the most riches you could ever dream of 💪
Lol
I thought the difference would be higher .
680m system is locked at 5 watts lower than 780m system as well.
the difference in 0.1% lows alone is worth it though.
I think the 780M is getting bottlenecked by RAM speed at this point.
Would love to see this same test but with 7200MHz RAM, now that the DDR5 prices have started to come down.
@@MidWitPride you're not getting 7200MHz SODIMMs anytime soon. At best, you could hope for LPDDR5/X soldered directly to the boards with the APU. Obviously, that would require a different PC from the one used in the video.
@@steph_on_yt correct right now the fastest is 6400 but that is still a fair bit better than 5600
Yeah I'm surprised at how small the difference is.
Someone who's not afraid to do some benchmarks even if the margins are tiny. It puts my mind at ease that I don't have to upgrade my 680M handheld any time soon. AMD have done a wonderful job at bringing gaming to the masses without breaking budget on a discrete gpu.
iGPU have come long way better! Well done AMD!
I would disagree. You could always play contemporary games with AMD iGPU on low settings. Also, these are best, most expensive models, and for that money you could buy low-mid range CPU and GPU that would run games better.
@@aleksazunjic9672 BS... Compare them to Intel's iGPU's. New iGPU's are way are better than old budget' 'flagmans' like GTX 750Ti or 1050Ti. Post author had a iGPU 'evolution' in mind, not a direct comparison between iGPU and a discrete card (it would be pointless).
@@aleksazunjic9672 But in terms of efficiency this tiny systems are very much better. I don't know any classic PC which reach this performance with only 50-60 w.
@@aleksazunjic9672 I agree, compared to PS5 APU, whose graphic performance is at par with a desktop RX6700 10 GB, a 780M iGPU really isn't that powerful. And far too expensive. AMD could deliver an adapted PS5 APU to desktop PC if they wanted to, couldn't they
fr
Honestly a bit disappointed with the uplift. Yes it's impressive, but the 680m was already impressive.
Yeah a great iGPU indeed
AMD doesnt have competition in the IGPU market. 6FPS IMO is laughable.
@@guilhermehenrique-zj5ttthe uplift in the average fps is ok, but the 1% and 0.1% lows are the real progress. Thase uplifts by themselfs are the real mvp since it means more stable and less noticable frame drops.
@@Gun10001 ^This
@@Gun10001 I wonder how much of the uplift in percentile lows is actually from the newer CPU or memory configuration (timings, latency, ranks). I somehow doubt that the GPU is the one boosting frametime consistency.
I just wanted to say thank-you so much for your inspiring work, especially concerning these Minisforum PCs.
Long story, short - I somewhat desperatly needed a little PC like this and your videos inspired me to get a UM690 with a Rysen 9. I know that my new PC has been surpassed by this new model but it wasn't released in time.
The fact that it can play some games with some compromises is enough for me 🙂.
3 weeks on:
- Still ebjoying ny new 690
- Got to tweak Windows 11 a bit 🙂
Thanks again!
Can you make a comparison with the old R5 2400g/3400g/5600g? Can't wait for these apu to come to desktop..
Small but nice improvement. RDNA3 being a testing architecture means the next gens could improve more
Very cool, interesting how well the 780m is doing in such a mini formfactor. Was wondering what the 780m would look like put up with a Ryzen 5600G or 5700G APU. Just to see how far the gap is between mobile chips and desktop APU's in 2023... Anyway hope you Enjoy.
And now that RAM prices are coming down too, i'd like to see some faster than 5600MHz RAM on these things tested out. You can get 32GB of 7200MHz DDR5 RAM for less than $200 these days.
@@MidWitPride I don't know what part of the world you're in, but SODIMM DDR5 seems to cap out at about 5600MHz here in the UK.
@@thedandyp OP was talking about future desktop CPUs, which won't be SODIMM only.
@@MidWitPridethe original poster very clearly states he wants to see the 780m up against the 5600g or 5700g. Specifically to see the difference between desktop and mobile chips, now, in 2023. No mention of any future CPU/APU at all.
Hmm, not bad, now I really want to see how these fare against older discrete GPUs like the 1050, 1050Ti and (2GB) 960. The latter is what I have in an old tower build, and yes, it is the 2GB variant.
It will obliterate them. My 680m on 7735HS is on par with my old R9 290x. You can also adjust VRAM in the settings, as vram is basically ram and DDR5 is very fast indeed (mine is 4800mhz). Impressive APUs.
@@TheThunderwars No its not, 780m is not even as fast as R9 290x
@@TheThunderwars Nah. 290X is faster than even RX 6400. It's closer to 6500 XT.
You should probably test it again when the official drivers from amd release for that igpu
Great review.
Your Forza skills have improved sooooooooooooo much lol.
Thank you for this. 10% is very much inline to what our knowledge of RNDA 3 improvement to be when looking at their desktop gpus. those 1% lows are incredible though, many games are much more playable.
I love your benchmark results graphic. How you display the results is fantastic
Thanks for the comparison!
Finally a decent comparison between 680m and 780m. I'm thinking of buying the minisforum EM680 or wait for the EM780, but the price difference will be approx $200. Do you think 16 gb is enough or should I get the 32 gb for $50 more?
16GB is enough but 32 will certainly give you more peace of mind, especially as the RAM is shared
igpu will take a chunk of RAM as VRAM, so instead of 16gb system ram you'll have smth like 12gb ram + 4gb VRAM. I'd advice going with 32gb ram, if possible
I'd go with 32. And then you can allocate 4 or 6GB for VRAM and still be left with plenty. Cutting 4 or 6gb off 16 won't leave you with a lot
get the 32GB remember you can easilly use 16GB for games alone... you need "VRAM" too... so get the 32GB and mark like 8GB for the APU and you are set...
These results are impressive to me from the point of any integrated graphics, but disappointing for the theorized uplift inperformance. I think retesting by increasing the usable vRAM limit from 4 GB to 8 GB a must. And a possible retest by upgrading the RAM kit in both 2 32 gigs and then having at least eight but up to 16 gigs of the ram in the test. I think that would be a good way to see the differences. and of course Do a comparison with the winner of these two to a 13th GEN Intel integrated graphics of any model
faster ram would help too... 5600 is slooooooooooooow
@@LiLBitsDK 5600 is the current fastest standard for DDR5 afaik
@@RM-el3gw there is DDR5-8000 out there already
Nicely done mate!
I really, really wanted this video. Thank you.
I started gaming on a A10-7800 APU. We've come so far in almost 10 years.
I would love a comparison of 780M based system, vs: a combination of standalone CPU + low-end graphic cards (like R6400?) 🤔
Unless the RX 6400 system had bare minimum R5 7600 or a 5800X3D the 780M system would wipe the floor with it.
Are you certain these tests were done in their native resolutions? I am seeing considerable aliasing!
Hmm, the 780M doesn't add as much performance compared to the 680M as I expected considering the sizable 400MHz clock boost. It does seem a bit more stable though, especially in GTA 5, the 680M's clock speed was all over the place, but the 780M held steady.
Would like to see a 780m Vs 1050/Ti
Excellent, im torn between the UM690 and UM790, this is extremely handy! Ta!!
lol literally just finished the last video and hit the homme button then boom new video what a follow up
RDNA3's jump in performance was overall disappointing compared to RDNA2's in discrete graphics cards and these APU results reflect that as well. The difference could easily be attributed to increased CPU grunt, power consumption and clock speed favoring the 780M.
This whole gen just screams "Skip Gen" to me, between the outrageous prices and disappointing performance bump
RDNA3 was supposed to be where they used chiplets to get more shaders on the card but they decided to only do 1 GCD for some reason. Really strange decision to throw away their advantage.
@@BonusCrook there are enough problems with RDNA3 already. 100w+ idle power consumption with multiple monitors, some of the VR games being broken, AV1 decode/encode sometimes not functioning correctly.. it's clear that the Radeon Drivers team didn't have enough time/resources to roll out proper support for RDNA3 even with single GCDs, 2 GCD designs would be a mess
I don't know if this could be interesting for you, but I would like to watch a comparison between moderns iGpu with popular Low Profile Gaming GPU.
Considering both can play triple A titles at 1080p low/medium settings is awesome.. If someone said integrated gpus would be capable of 1080p gaming a few years ago people would think its a dream, but now its very possible.. If the person doesn't care about playing at 900p or even 720p then its definitely possible to play at 60fps high.. Sadly nowadays 1080p is considered minimum, but having fun is the most important, even if you have to sacrifice graphics and resolution to have playable fps.
and it will be even more possible thanks to fsr / dlss stuff
So i have the beelink ser 6 pro 7735HS with 680M. I also got it only for 350 BUCK! But i was wondering if i should upgrade to minisforum with 780M graphic. What do you think should i upgrade or am i good to go for now.
Would love to see you test them in the future when faster ram is available, because i really think 5600mhz ddr5 ram is kind of a bottleneck(even the 680m will be improved)
Came to say this. The way I understand, these iGPUs really benefit for faster RAMs, so there is probably a few extra %'s of performance to be had there...
@@rhobson it honestly could be more than a few %'s. The 6800u could already perform better with 6400mhz lpddr5 ram and it was a decent uplift. The 780m isnt the double the performance for sure but with 7400mhz lpddr5x ram it will without a doubt perform much better than shown in the video. And also increasing the access to 8gb would help a lot in vram intensive games,which almost all modern games are, but with 16gb of ram it not a good idea(32gb would easily allow it).
I like those small Minisforum machines and I consider buying one for work mainly to lower power consumption. My main PC needs 65W in idle with one monitor and 95W for two monitors. It's just a waste of energy and in summer makes my room even hotter.
Really thinking about getting a mini pc with the 780m, I'd like to use it as some sort of emulation system for older game consoles.
It* remarkable that the these APUs (particularly the 7940hs w/780m) are in Laptops now as well, Gaming laptops at that, where they can now allow these behemoths that once can only last for about 3 to 4 hours at most even with just playing indie stuff or web browsing, to last a whopping ten hours without even touching discrete graphics at all!
Just FYI laptops have used some form of dual graphics way before this CPU came by man, I had GTX cards along with Intel HD some 15 years ago....given it wouldn't stay on for as long as today's laptop can.
Looks like a huge bandwidth limitation of the RAM.
I hope you would do the comparison between 780M and new AI 890M 🙌💫
Is the memory in the test 4800mhz or 6400mhz? Since the 6900HX can also handle 6400 mhz memories.
According to his intro screen, he was running both at 5600MT, but I would have liked to see that in the overlay. I have only been able to convince my 690S to run the memory at 5200MT (default is 4800MT), while the 790Pro defaults to 5600MT (providing you have memory that supports that speed).
How much memory were you using, and how much did you allocate to VRAM?
I remember using a 2200g with a Vega 8 igpu in it. Actually, makes me play games at low or medium pretty well. AMD surely has come a long way
I have a 6850U with the 680M in a 14" thinkpad and I'd predict in more realistic use cases in laptops there would be very little difference, under 28W STAPM I rarely hit the maximum 2200MHz GPU clock.
The 780M by design and characteristics is at best 10% faster than the 680M. In the new APUs, the cpu is a bigger upgrade imho and I believe that is likely what you are seeing on the increased 1% lows. Quick way to test this is to OC the 6800H to try and match 7940Hs clocks. Or just match the clocks if that is not possible.
It does raise a question about price to performance as you can get a 6800H system for about 350 pounds whereas you are looking at at least 500 for a 7840HS and more for a 7940 system.
@@missrochelle1126the new ones yeah. I mean, 500 quid for a whole computer is not that much especially at this level of performance I mean, you can play modern AAA games with upscallers at 1080p looking gorgeous. They pack a lot of performance.
@@HeldermaiorI can build a 6700xt system with the same amount of money
Do you think AMD 780 is better than RTX 500 ADA? I see that increases the price by a lot when selecting this option. I am looking to do 2D, 3D work
I was expecting a lot more because of the massive clock boost and DDR5 memory
well both used the same memorry so only the clock and architecture makes a difference :)
What are these emojis can't seem to find them on my keyboard
1. Do igpus work at full power when a laptop is not plugged in?
2. Are there any good 780m laptops with long lasting batteries?
How does the 7840U allocates VRAM for the 780? Is it automatic?
i'm confused about the vram usage in rdr2. why is 780m system using literally half the vram of 680m system?
Is there any chance that you try what the 780M/680M can do in WQHD?
Minisforum has the ryzen 7 7735HS and 680m UM773 for $539 and that’s a steal for a very compact pc with a gaming performance near to dedicated gpus from the yonder years lol
are the vram numbers also filling up ram space ?
There is little to no performance improvement from rdna2 to rdna3 per (core*frequency). Same story with the rx6600 and the rx7600.
Hyped for the 880m (or even the 990m) it'll be amazing for sure!
Is the Radeon 680M way better than Nvidia T500 for ~lighter games like DayZ ? I purchased a Thinkpad laptop with T500 and I can still cancel it to get another Thinkpad with Radeon 680M ($400 more expensive).
what i do is low to medium preset then setting the texture quality to high or ultra it looks way better then low textures with less fps drop you can test in a video with few of your cards if you want
rx 6700xt 12gb is card im using
Should these to able to take on a 1050 or 1650 GTX?
i wish you had unbottlenecked systems by having 2x16gb memory kits compared to only 2x8 (or is it 1x16?)
Is there any info on how this compares to a 5600g?
finished BC2077 3 times, you just made me hungry for another run
How would Arma3/4 or DayZ perform with these integrated graphics options?
What happens if you install 32 GB of ram and give the GPU something like 8 GB of VRAM? Just thinking if the VRAM held them back and you would see bigger differences
Would love to see beelink gtr7 pro vs minisforum um790 pro
Either it's the TPD throttle on the 780M _(on that board),_ or just it's not worth it.
Because, taking in mind the new generation _(mind the 2800MHz vs 2400MHZ!)_ - the fps difference just doesn't add up to an expected value, especially considering how triumphant were the "new king" claims for the 7xxM presented by AMD.
Funily enough i own both these processors if you count the R1 Extreme. But since the laptop that has the 6800h in it has a 3050 ti the most I can tell you is the 680m isn't enough to render 2400p HDR content to the oled monitor its attached to. Considering I already have to manually limit the TDP of the 6800 so it doesn't burn itself alive at 100 degrees, I cant imagine adding to that heatload with the igpu.
BUT ... if u have used 2x16 GB ddr5 5600 or higer speed? did it count more fps?
Curious if you could run the same test with 32GB of RAM to see what the difference would be
Have you ever tried one of the more milsim type of games on these mini pc's? Like ArmA 3/Reforger, Squad, Hell Let Loose? Like to hear from you!
I'm here for exactly that. I wanna know as well
The newest version? Maybe. These meet the minimum CPU but seem borderline for GPU - rx570...that's about where these are...
It's funny because the Ryzen 7600's iGPU in the IO die overclocks to 2800 Mhz @ 1.3V before hitting a power wall, too. Of course that's only 2 CU but that just makes the overclock sweeter since it has a direct and tangible impact on frame rates.
It's interesting how they're both power limited but also manage to reach full GPU utilization.
These 680m or 780m are better than those included in 7000 desktop series (7600,7700,7900) ?
they are nowhere near these cards, but its still amazing what they offer in such a small form factor
Can it do any ai work for those playing and learning ai?
how does a 5700G compare as well ?
As many have pointed out a comparison with same power limits/frequency would most likely show the results to be equal. Which does bring into question whether the 780M is a "true" RDNA3 a.k.a. generation change.
Regardless it is very uplifting to see that both minipcs as well as "budget gaming" can utilize these dGPUs for 1080p even with new games.
the 780m draws a few watts extra because the cpu cores clock slightly higher, the gpu TDP is pretty much the same and for that same power it scales to 2800mhz instead of 2400mhz which is mostly where the performance comes from.
But scaling a 680m to 2800mhz wouldnt get the same results as RDNA3 has a different shader layout (yeah both have 12CU but that doesnt tell you anything), RDNA2 has 2x SIMD32 units per WGP but RDNA3 has 4x SIMD32 units per WGP to more efficiently distribute work within the core as RDNA3 of the same shader count has half the WGP's, this was done largely to scale to bigger cards like 7900xtx but it still scales.
I'm still using the 4650G, not the best obviously, but I wonder how much the difference between mine and the couple shown in the video. 🤔
What about the noise? What's the noise like on these units?
41dB on the 680m not sure about 780m tho probably quieter? Since it's more efficient.
I wonder when we'll get some desktop apus from amd
There was no difference on the Witcher game?
That is amazing for a mobile APU. Imagine if we get a desktop version that's not scaled back, we're talking Xbox Series S power, probably.
Yeah I’m looking forward to a new desktop APU. Long overdue!
Noob question
If i play a game capped at 60(not an FPS), will it make any difference vs if i uncap the fps? What's the purpose of a game going over 60?
Precision is one of the more useful advantages of a higher framerate.
If it takes 1 second to rotate 360 degrees at 60 frames per second then each step of the rotation is 6 degrees. If you're getting 120 frames per second then each step is only 3 degrees.
It doesnt take a second to rotate through a full 360 degrees for most games of course, and the numbers I used in the example are picked at random, but the theory behind it still stands.
It also applies still when using a lower refresh screen, even when the framerate exceeds the screen refresh, so there are advantages to the framerate exceeding the screen refresh rate.
How does this compare to something like an RX 580 or RTX 3050?
Not well i think
perfect video
Will the 1080m be like a gtx 1080?
Difference not that big which is kinda sad, but on other hand older models gonna cost less now and you can grab something quite fancy during Black Friday etc.
I just want an APU with the performance of either of these for desktop diy.
I had considered buying both of these at one point in time. But after seeing the performance I have to say I'm very disappointed. I expected GTX 1650 performance like the products had claimed. Instead we got GT 1630 level of performance.
arent these igpus faster than 1650 ?
Wonder how much of a difference the 7940hs vs the 6900hx is making here. The cpu is also much more powerful here. So that isn’t even being factored in the difference between the 680m and 780m. The 780m might literally be a less than a. 5% upgrade
I wonder how this compare to the last desktop APU's AMD gave us aka 5700G and 5600G and 4600G etc.
i7 8809G is known as the strongest core display, and I am very curious about their previous competition results.
You're using overclocked RAM for the 680M, the native support is 4800MTs. Only 780/Ryzen 7000 Mobile's is 5600MTs. But once more, a much better setup would have been LPDDR5-6400 for the 680M and LPDDR5X-7500 for the 780M, but I know these boxes come in with non-soldereded/LPDDR RAM.
LPDDR5 is slower than equivalent DDR5. Don't just look at frequency it's not equivalent. Memory is 2 things, frequency and timings. If you want to compare, LPDDR5 @ 6400 is between DDR5 5200-5600
Great video, as always!❤
Thanks again!
You dont have any latest graphics driver for 780m. we should wait and see.
Gota add the 880m and 809m in anew vid
The biggest question is if we paired the 6900hs with 6000 mhz ram and the 7940hs with 4800mhz ram what sort of difference we looking at
It's 800 Euro with 32GB and a 1TB SSD. I don't get the appeal of it. It's laptop parts, but not really portable and costs a lot. A small and efficient desktop PC with a more powerful discrete GPU would cost roughly the same and have better gaming performance.
I got a 680m in my miniPC ... Finally 780M is just a very little upgrade
expected more of an upgrade, nvidia mobile gpus performance boost have been great latelly, unlike their desktop parts
want to see both igpu overclocked, ram overclocked, also 32Go of ram..
*We must all remember*
*If a desktop or laptop iGPU outperforms their discrete GPUs, then they lose sales on those cards (as they'd be obsolete). This small leap was done on purpose. The 780m already performs as well as a GTX 1660* 😎
This completely false, 780m is slower, in general by a decent margin, to a 1060. 1660 is substantially better.