I totally concur.
Building a new PC for Elite: Dangerous Odyssey (hardware discussion)
- Thread starter Alec Turner
- Start date
I totally concur.
You sound like a apple adept. Upgrade-ability and customization are the greatest strengths of PC building, and millions of PC enthusiasts around the world will never give up that feature despite apple wanting to take it all away.
I used to build PC's for a living, about the only thing that got re-used from old PC's was, sometimes, the case. Sure customisation is great, but if you build a decent PC to start with by the time you decide to "upgrade" none of the components will be useful in a new motherboard, it's almost always a full replace, unless of course you are one of those PC users who replace parts every year or so. Selling ordinary users on "upradeability" is always a mistake, unless they can build it themselves, and can afford to buy new components and parts on an ongoing basis fairly rapidly it's never going to happen. Of course we have the "ship of theseus" factor then, if you have replaced every part is it still the same PC being upgraded or indeed have you just built a new PC over a couple of years. That of course is a costly way to do it as well, because buying bleeding edge components as soon as they come out rarely saves money in the long term.
there's a huge difference between being a PC seller and being a PC enthusiast. The latter are an army not to be underestimated because if they were, Apple and their ilk would've gotten their way a long time ago. You, on the other hand, are the Apple-like PC seller (former or not) who does want to take away all the benefits of flexible PC platform simply because it's not profitable enough for your and your business.
Last edited:
That's bizarre, nearly everything you said then was entirely wrong. I said I built PC's, didn't say I sold boxes, I custom built them for people whose PC was so old that literally nothing in them was at all useful, that's your average person, that's 99% of the population in fact. Box sellers, Harvey Norman et al, aren't PC builders, they are box sellers, there's a difference, I suggest you understand that.
and yet you parrot Apple-like platitudes about PCs. Sorry but you don't appear to be any different.
i had an msi once back in the days of win 95
which broke 4 days after warranty expired
there only good product is afterburner imo
which broke 4 days after warranty expired
there only good product is afterburner imo
Last edited:
I've bricked two MSI B550 boards (of different models) badly enough that the BIOS flashback feature couldn't recover them. Since they used those annoying UFDFPN8 EEPROM packages I couldn't get my programmer clips to engage with and I was disinclined to desolder and resolder (or make an adapter for the header on the boards) them just to program them, I sent the boards back to MSI, under warranty, and got them reflashed with no problems (and at no charge). Was actually a fairly smooth experience. Not that I'm doubting any one else's personal experiences with them.
Less anecdotally, they have been pretty careless with customer info and their own security, which is always concerning.
Odd, as I've never owned any Apple product.
I've built hundreds of PCs and I average about two a year for my personal use. I also frequently reconfigure and upgrade my existing systems. It's a major hobby of mine.
However, being a PC enthusiast (which automatically makes me an outlier among PC users) doesn't blind me to the fact that a custom build isn't always going to be the most optimal way to get a system that meets every set of constraints. For most enthusiasts the build process in and of itself is a big part of the appeal, but that doesn't mean it's going to be the pragmatic choice.
If someone wants a hobby project, that's one thing, but the idea that a portable, ~100w system, with an integrated UPS, is somehow better served by the hypothetical upgradability of a hand-built system than an off-the-shelf laptop is something I believe I can objectively disprove.
Apple is hardly the only anti-consumer, anti-customization, anti-repair, anti-third party, planned obsolescence, consumer electronics corporation out there...but Apple doesn't have much of anything to do with anything here.
We are a tiny minority with a barely outsize market influence and the idea that hobbyists like us are the reason there are still replaceable components is insane. Enterprise clients and OEMs are why these things exist.
Apple and their ilk already got their way, and they can ignore the existence of hardware enthusiasts because such enthusiasts are a tiny drop in a very large bucket of potential revenue. Apple is a 3.4 trillion dollar company, which is probably multiple orders of magnitude larger than the entire PC enthusiast and enthusiast-adjacent market. Companies like Apple have been largely successful in turning computers into disposable devices-as-a-service, with pro-consumer legislation barely presenting a speed bump to their robber baronies. Even core hardware manufacturing has long since devolved into syndicates that are in essentially permanent states of collusion and price fixing, because even where such behavior is illegal the penalties dwarf the profits.
You might want to look into Framework laptops. It's a neat idea.
TechPowerUp
Leading tech publication with fast news, thorough reviews and a strong community.www.techpowerup.com
There are many interesting details about CPU/GPU types in the link above and I'm often browsing this site to see if something interesting comes out.
I was laptop user for most of the time, portability was great but limitations were too serious for me. Now I can replace any part I want whenever I want. The main issue for me is a case since buying one I want is nearly impossible- it has to be done by me- and my hands that require more training to achieve desirable result. My ideal PC is in the form of suitcase but there are some technical limitations I have to solve, like building my own custom PSU that can work with LiFePo4 pack (that have to be done by someone knowing more than me about electronics, charging curves for batteries I want, etc.). Right now I'm using normal, mini tower type case that can also fit 12Ah AGM lead battery but that is only temporary until I have time and resources to experiment more.
It might, however, not meet your TDP and sound requirements.
(deleted post)
I think that comparison works when you are happy to make changes in a big bang approach, ie. like Alex at the start of this post, and if you go down the laptop route, you have no choice.
However my experience was different, I started with a slim and light i5 laptop that would play the original pre-Horizons ED (I got hooked around the 1.2 Wings update).
Next I got a pair of 1080p screens for WFH and of course had to use them for ED, so started to suffer when asking the latop to run 3840x1080.
My first PC build was solely for ED. AMD fx 4300, Nvidia 960, 8GB DDR3 and a 500w PSU. I set myself a £400 budget at the time (any more and I could have just bought an XBOX/PS4).
This was great until Horizons dropped (and also I had an eye towards VR), so my next purchase was to upgrade the 960 to a 1070
Next was the awkward one, and probably where most the arguments about full system replacements come from. AMD had moved platform, so I could drop a couple of year out of date FX8300 or upgrade CPU,MB and to DDR4. I flipped to intel and an i5-8400, and smaller motherboard.
At this point I was still in the same case, and had same PSU after 3 yrs. If I had been on laptops, I would probably have been spending lots more to get to run VR, and again for Odyssey.
Since then the only performance upgrades really was the 3080FE, but that would have needed a PSU upgrade had I not already bought one to fit the smaller case (and because its an SF600 Platinum, it has coped with to 3080 power spikes without incident yet). I also got a cheap 32GB of memory on last years Prime day,
- https://uk.pcpartpicker.com/b/Q7fH99 (This was the final configuration and had to be run without glass panels. The Shift looked pretty but ran hot)
However, I've had all sorts of other upgrades because I discovered I enjoyed building/tinkering with PC hardware. Stuff that you just cannot do with laptops
I'm currently looking towards another CPU upgrade, which also (once again) means CPU+MB+Memory. In some ways I've been really lucky to get a good enough experience on a 6yr old low end i5, but that's the limit now (Horizon Forbidden West on 4K Ultra, but with occasional FPS dips) (Titan bombing in Open with all hundreds of Thargs with occaisional stutters). A 14600K was calling, but I'm now hanging on to see how both next gens look.
Last edited:
Yup, I think I personally am in the Veronica/Morbad camp on upgradeability. It's seems like a good idea but in practice I always end up making do with the PC I've got, possibly for too long, then finding the motherboard or something else won't support a meaningful upgrade, and then just deciding to get a whole new PC. The only upgrades I've ever done are to add more memory and to change from a GTX 970 to a 1080 when I got VR. I appreciate others will upgrade regularly and enjoy spec'ing a PC with that in mind.
My prior pc was an i5-2500k cpu, with a radeon 6850 gpu. I never upgraded the cpu (just overclocked it), but the gpu was upgraded to a 1070 gtx. That system is stilling running today as a backup pc, it is no longer my main system.
My current main pc started as a Amd am4 r5 3600 with a 1660 Super (due to crypto prices of gpus). A few years later I have the same mobo, ram, case, and power supply, but a 5800x3d cpu and an rx 6800 gpu.
I usually upgrade my pc's, which is why I do not like laptops. As laptops are pretty much non upgradeable and usually they cost as much or more than a comparable desktop initially. And for me at least PC upgrades (cpu, gpu, ram) are easy.
My current main pc started as a Amd am4 r5 3600 with a 1660 Super (due to crypto prices of gpus). A few years later I have the same mobo, ram, case, and power supply, but a 5800x3d cpu and an rx 6800 gpu.
I usually upgrade my pc's, which is why I do not like laptops. As laptops are pretty much non upgradeable and usually they cost as much or more than a comparable desktop initially. And for me at least PC upgrades (cpu, gpu, ram) are easy.
This is what I did.
I started with an R5 3600 in 2019 with 16GB of DDR4-3200 and an RTX 2060. The system stayed the same until I handed it over to my son, who has since upgraded the RAM, CPU and GPU to a 5700X, 32GB of DDR4-3600 with tight timings and a 3070 Ti, which he got when I upgraded to the 4070 Super.
I started a new build a couple of years back, which has resulted in the specs in my sig. I plan to keep it for at least another two years, hopefully more. I could have kept the original motherboard, but my boy needed a new PC. My other son has an evolution of an R5 2600 build from 2018 using an Asrock Mini-ITX in an appropriately small case. It's now running a 5800X with 16GB of DDR4-3200 and a 3060.
Cascading parts down and upgrading the platform when needed was the cheapest solution to having a family who all play games on PC.
The total longevity of the Asrock Mini-ITX motherboard has already exceeded 6 years and looks good for another two at least. By that time my boys will be in their twenties and can buy their own PCs.
The main factor that makes this more or less viable is platform support. AMD's policy of continuing to support AM4 for 8 years has been very helpful in my situation. YMMV, of course.
Last edited:
Don't get me wrong, I'm not against upgrading existing systems, nor am I against factoring in upgradability when building them. I'm against blindly following any doctrine, especially when it contravenes one's overarching goals, or ignores relevant context.
Here's an example of an upgrade path for one of the systems I was using when I was playing Elite: Dangerous 40 hours a week (a system I still use, but which is about to be partially cannibalized/downgraded in favor of a new build):
CPU -- 3900X -> 5800X -> 5800X3D
Memory -- Hynix CJR -> Samsung B-die
GPU -- RX 6800 XT -> RTX 4090
With Odyssey performance being so profoundly memory subsystem sensitive, going from an R9 3900X to a 5800X, then upgrading the memory, then swapping the CPU again for a 5800X3D produced major performance improvements at each step. The final build, with non-GPU limited settings, gets more the twice the frame rate in large settlement surface conflict zones, compared to the original build. The RTX 4090 was added a bit later, after the path Odyssey was taking had significantly dulled my enthusiasm for the game and I found myself with time to play some other titles, but was also very useful in Odyssey for mitigating those jaggies I find so annoying.
GPU power and cooling constraints were an issue with a (barely) SFF system and the RTX 4090 upgrade, but the cost of adapting to that (new case, PSU, and cooling system) was a small fraction of the cost of the GPU itself. I just recycled the initial case and PSU for the HTPC build; bought an adequate PSU that was on sale (a Seasonic Focus Platinum 850); paired it with the cheapest, most compact, and least offensive case compromise (ASUS AP201) that could hold the RTX 4090 and it's cooling; then build a budget custom loop sufficient for the GPU (which can be way less expensive than most people seem to assume).
Other critical context here was that there were no meaningful platform upgrades possible at the time. LGA-1700 would barely have improved performance and is not particularly SFF friendly (I can do an RTX 4090 or I can do an i7/i9...both would start to overwhelm cooling and/or trip breakers). AM5 hadn't launched yet, and wouldn't be decisively better until Raphael-X. AM4 was also an unusually long lived platform. Plus I had a use for the parts I was replacing (that 5800X and RX 6800 XT are in my HTPC).
These upgrades were relatively easy choices that weren't barred or rendered impractical by other constraints. Such is not always the case. Indeed, such is often not the case.
Some constraints close off most practical upgrade avenues, and that was the context for my laptop recommendation in prior posts. What low-profile and low-power GPU is one going to upgrade to that is going to beat what one could have had in a mid-range gaming laptop from the start? Why bother building a custom UPS to fit in a mini-PC when that's what the battery in a laptop already is? What Intel platform is going to accept a meaningful CPU upgrade, if one didn't waste money handicaping one's self by low-balling the original part? How are you going to match the power efficiency of a single board computer, made from parts binned for efficiency, with desktop components, that, at the very least, have multiple PCBs and power conversion systems setting a much higher energy floor? These are semi-rehtorical questions. The only way to beat a sensibly chosen laptop, over any arbitrary span of time, under Ameeleean2's implicit constraints, would be to spend way more money, or invest a huge amount of extra effort adapting/fabricating custom parts to fill a niche that a laptop already fills.
A laptop isn't very upgradable, but upgradeability is a means to an end, not an end in and of itself. If the journey is the goal, that's great...no one can put a price on the experience except the one experiencing it. However, if the goal is to have a computer that does what the afforementioned custom build will do, the laptop route would get there, and stay there longer, with better performance, lower power, and probably less money.
Here's an example of an upgrade path for one of the systems I was using when I was playing Elite: Dangerous 40 hours a week (a system I still use, but which is about to be partially cannibalized/downgraded in favor of a new build):
CPU -- 3900X -> 5800X -> 5800X3D
Memory -- Hynix CJR -> Samsung B-die
GPU -- RX 6800 XT -> RTX 4090
With Odyssey performance being so profoundly memory subsystem sensitive, going from an R9 3900X to a 5800X, then upgrading the memory, then swapping the CPU again for a 5800X3D produced major performance improvements at each step. The final build, with non-GPU limited settings, gets more the twice the frame rate in large settlement surface conflict zones, compared to the original build. The RTX 4090 was added a bit later, after the path Odyssey was taking had significantly dulled my enthusiasm for the game and I found myself with time to play some other titles, but was also very useful in Odyssey for mitigating those jaggies I find so annoying.
GPU power and cooling constraints were an issue with a (barely) SFF system and the RTX 4090 upgrade, but the cost of adapting to that (new case, PSU, and cooling system) was a small fraction of the cost of the GPU itself. I just recycled the initial case and PSU for the HTPC build; bought an adequate PSU that was on sale (a Seasonic Focus Platinum 850); paired it with the cheapest, most compact, and least offensive case compromise (ASUS AP201) that could hold the RTX 4090 and it's cooling; then build a budget custom loop sufficient for the GPU (which can be way less expensive than most people seem to assume).
Other critical context here was that there were no meaningful platform upgrades possible at the time. LGA-1700 would barely have improved performance and is not particularly SFF friendly (I can do an RTX 4090 or I can do an i7/i9...both would start to overwhelm cooling and/or trip breakers). AM5 hadn't launched yet, and wouldn't be decisively better until Raphael-X. AM4 was also an unusually long lived platform. Plus I had a use for the parts I was replacing (that 5800X and RX 6800 XT are in my HTPC).
These upgrades were relatively easy choices that weren't barred or rendered impractical by other constraints. Such is not always the case. Indeed, such is often not the case.
Some constraints close off most practical upgrade avenues, and that was the context for my laptop recommendation in prior posts. What low-profile and low-power GPU is one going to upgrade to that is going to beat what one could have had in a mid-range gaming laptop from the start? Why bother building a custom UPS to fit in a mini-PC when that's what the battery in a laptop already is? What Intel platform is going to accept a meaningful CPU upgrade, if one didn't waste money handicaping one's self by low-balling the original part? How are you going to match the power efficiency of a single board computer, made from parts binned for efficiency, with desktop components, that, at the very least, have multiple PCBs and power conversion systems setting a much higher energy floor? These are semi-rehtorical questions. The only way to beat a sensibly chosen laptop, over any arbitrary span of time, under Ameeleean2's implicit constraints, would be to spend way more money, or invest a huge amount of extra effort adapting/fabricating custom parts to fill a niche that a laptop already fills.
A laptop isn't very upgradable, but upgradeability is a means to an end, not an end in and of itself. If the journey is the goal, that's great...no one can put a price on the experience except the one experiencing it. However, if the goal is to have a computer that does what the afforementioned custom build will do, the laptop route would get there, and stay there longer, with better performance, lower power, and probably less money.
Hi All
As usual an interesting read.
I personally find that buying a new graphics card is the most difficult and frustrating part of upgrading / building a computer. I think that's mainly because it's usually the most expensive single part compared to the other individual components. Parting with that 'block' of money in one go so to speak is not in one sense enjoyable.....
Jack
As usual an interesting read.
I personally find that buying a new graphics card is the most difficult and frustrating part of upgrading / building a computer. I think that's mainly because it's usually the most expensive single part compared to the other individual components. Parting with that 'block' of money in one go so to speak is not in one sense enjoyable.
....Jack
For me hardest part is finding out that hardware of my choice has little tiny detail, that usually is mentioned in manual but omitted on web page general product description. Let me give an example. When I was upgrading my PC with new graphic with card I have carefully chosen (not carefully enough apparently) it has turned out that my CPU and MB is 2 generation behind rebar support (there were ways to enable it but too complicated and too risky for me), so it was either to return product or replace MB+CPU. I have chosen later so I could build extra PC for my son- he was old enough to have his own. The funny thing is that Inter Arc without rebar has terrible performance and my old system with just nVidia RTX 3050 is outperforming in games my current system, but there were different requirements for both PCs and in both cases I'm satisfied with the result. I was also dissapointed with Intel approach to CPU design (more cores, higher power requirements, higher TDP, lower base core clock, the same transistor size) so I have ended up opting for lowest generation CPU that matches my criteria and has the same or lower transistor size, therefore chosing Intel i10-10100T as replacement for intel i5-7500T- imho there is no point in naming CPU as next generation when it uses the same transistor size as its predecessor, as for higher number of cores I would only need more than 4 if I want to run more than one VM on my hardware (which is why I have 32GB of memory btw).Hi All
As usual an interesting read.
I personally find that buying a new graphics card is the most difficult and frustrating part of upgrading / building a computer. I think that's mainly because it's usually the most expensive single part compared to the other individual components. Parting with that 'block' of money in one go so to speak is not in one sense enjoyable.
Jack
This is also why I prefer PC over laptop, if I wanted to do the same thing while owning laptop I would be simply forced to buy new one.
Few people like parting with money. Just set a budget and get the best option that fits.
But yes, GPU is and should be the most expensive component of any system where playing modern games is a serious consideration. When building a gaming-oriented system, my rule of thumb is that the GPU should be half the price of the build, except at the very high-end, when it should be closer to two-thirds of the budget.
I frequently see people blow huge amounts of money 'balancing' the rest of their system against a high-end video card, which is usually a completely senseless approach.
Naming conventions are all marketing, but there is more to manufacturing process than feature size and more to generational CPU changes than changes in manufacturing process. It is true that 6th through 10th generation Intel desktop CPUs were mostly minor iterations on Skylake. 11th generation desktop parts were a major architectural update, but were backported to a more mature (and less efficient) process to meet clock targets. Throughout that period Intel reserved most of their efficiency improvements for mobile SKUs and quickly fell behind AMD in desktop efficiency.
AMD has had the desktop performance/watt edge since Zen+ (Ryzen 2000) and Intel, assuming their management doesn't run the company into the ground in the meantime, probably won't regain competitiveness in this regard until Panther Lake (16th ot 17th gen).
Things have been very different in the mobile space. AMD really only became competitive these last few product cycles.
Careful selection of the initial system should obliviate the need to do something similar with a laptop. If one knows that swapping the CPU and GPU are out of the question, then one can choose to get something that will be capable enough to last as long as one needs it to last from the get go.
For example, in late 2018 I bought a discount (as it was a last gen model at the time) gaming laptop with an i7-7700HQ and a GTX 1070 Mobile for a friend's birthday. With the sales going on, it cost about a thousand bucks at the time, but is still as fast as a system with an i3-10100T and a desktop RTX 3050.
Today, a similar purchasing strategy would get one an i7-13700H or R9 7940HS paired with a mobile RTX 4060 16GB or RTX 4070 16GB (both of which are considerably slower than their closest desktop equivalents, but will be more than sufficient for 1080p gaming for years to come). Though waiting until early next year for Strix Halo and other options to drive prices down further could be wise.
I was thinking of using laptop MB in my custom PC since it has some features I want:
1. Small size
2. Mobile CPU type with better power management (it is easier to get new gen CPU with lower wattage and TDP).
3. Built-in UPS (since it works with lithium type battery from the start)
4. Screen matrix that can also be placed in suitcase cover.
but there are drawbacks too:
1. Non-standard size, since it is designed for specific case already (this can be amended but new MB=new mounting in case, screw holes etc.).
2. Non-replaceable CPU with custom cooling system, which is usually quite loud when system is under stress (this probably can be amended by water cooling system).
3. Built-in UPS that can be used exclusively with lithium battery having proper voltage ranges
4. Inability to use low profile (or normal) PCIE GPU card, unless MB has thunderbolt support which can be used as PCIE-PCI adapter, not to mention power circuit paths and cables might be unable to handle higher amperage (meaning extra GPU dedicated PSU). I've been also reading a lots of complains about this solution.
100W limitation has also other more technical reason for me (beside work time on battery), which is safety limit for cable cross section area size (each 10A equals to about 1.5mm2, but this also depends from conductive material type), battery voltage when under heavy load and when idle state (for example voltage on typical car lead-acid battery during engine start can drop even to ~6V)- for my PC under full load battery voltage was between 10V-11V and lower voltage=higher amperage=higher cable cross section area size. That also applies to printed circuit paths. I recall seeing ppl complaining about burned power connectors for high-end GPU cards which is direct result of breaking this rule or or staying too close to safety limits so yes, this is important for me.
In the end I have chosen mini-ITX MB format as a base, it is small (17mm x 17mm), has one PCIE slot (just what I need for one LP GPU card), usually built-in WiFi extension card and can be used with pico-PSU (preferably with wide input Voltage range) to save even more space. Right now I'm using a type of open PSU with UPS capability but it can work only with lead acid battery type (since it was designed to be used in cars). Not to mention this setup is more effcient in power conversion of whatever is in power socket (for me that's 220-240V) for DC power used in PC, than typical PC PSU. I can also use directly power from a car battery or any other source (therefore bypassing extra PSU with UPS), providing it is in input range of wide range Pico-PSU I have atm.
As for gaming all is good when I have decent fps in few games I like. I remember even playing ED horizons on integrated GPU- something obviousy not doable with oddysey DLC. If someone is curious total power usage under full stress was around 55W (all power conversion losses, monitor and peripherials included).
I had to add dedicated GPU card to handle some games (ED Odyssey included) so I have ended up with higher power consumption. I just wish some nVidia chips meant for laptops were available for desktop PCIE slots, that would be perfect for me.
1. Small size
2. Mobile CPU type with better power management (it is easier to get new gen CPU with lower wattage and TDP).
3. Built-in UPS (since it works with lithium type battery from the start)
4. Screen matrix that can also be placed in suitcase cover.
but there are drawbacks too:
1. Non-standard size, since it is designed for specific case already (this can be amended but new MB=new mounting in case, screw holes etc.).
2. Non-replaceable CPU with custom cooling system, which is usually quite loud when system is under stress (this probably can be amended by water cooling system).
3. Built-in UPS that can be used exclusively with lithium battery having proper voltage ranges
4. Inability to use low profile (or normal) PCIE GPU card, unless MB has thunderbolt support which can be used as PCIE-PCI adapter, not to mention power circuit paths and cables might be unable to handle higher amperage (meaning extra GPU dedicated PSU). I've been also reading a lots of complains about this solution.
100W limitation has also other more technical reason for me (beside work time on battery), which is safety limit for cable cross section area size (each 10A equals to about 1.5mm2, but this also depends from conductive material type), battery voltage when under heavy load and when idle state (for example voltage on typical car lead-acid battery during engine start can drop even to ~6V)- for my PC under full load battery voltage was between 10V-11V and lower voltage=higher amperage=higher cable cross section area size. That also applies to printed circuit paths. I recall seeing ppl complaining about burned power connectors for high-end GPU cards which is direct result of breaking this rule or or staying too close to safety limits so yes, this is important for me.
In the end I have chosen mini-ITX MB format as a base, it is small (17mm x 17mm), has one PCIE slot (just what I need for one LP GPU card), usually built-in WiFi extension card and can be used with pico-PSU (preferably with wide input Voltage range) to save even more space. Right now I'm using a type of open PSU with UPS capability but it can work only with lead acid battery type (since it was designed to be used in cars). Not to mention this setup is more effcient in power conversion of whatever is in power socket (for me that's 220-240V) for DC power used in PC, than typical PC PSU. I can also use directly power from a car battery or any other source (therefore bypassing extra PSU with UPS), providing it is in input range of wide range Pico-PSU I have atm.
As for gaming all is good when I have decent fps in few games I like. I remember even playing ED horizons on integrated GPU- something obviousy not doable with oddysey DLC. If someone is curious total power usage under full stress was around 55W (all power conversion losses, monitor and peripherials included).
I had to add dedicated GPU card to handle some games (ED Odyssey included) so I have ended up with higher power consumption. I just wish some nVidia chips meant for laptops were available for desktop PCIE slots, that would be perfect for me.
Last edited:
If you can find a part that uses the same or similar die flavor (much easier to do with AMD than with Intel currently, as Intel's manufacturing issues have forced them segregate product lines more), you can erase most of the differential in power consumption because power consumption is largely an arbitrary thing.
A large part of whatever power differential is left is at the board/platform level because most quality desktop boards are designed to support a much wider range of much higher power parts, and the boards that aren't are typically inefficient junk.
This can be supplemented with, or replaced by, an external (typically USB-C) battery.
Even mid-range laptops are starting to include multiple M.2 slots. An M.2 M-key slot is just a PCI-E x4 slot and if you can get by with four lanes (many lower-end GPUs only have four or eight lanes or won't suffer from less, if they have sufficient VRAM) it's not difficult to attach a GPU to an M.2 M-key slot with off the shelf passive adapters.
By and large, they are. Minor binning differences aside, power consumption here is just as arbitrary as with CPUs. An RTX 4060 Max-Q that might be in a budget gaming laptop is the same chip (AD107 in this case) as the desktop RTX 4060, it's just clocked significantly lower.
The problem, again, is the board losses, which still exist even for things like MXM adapter cards. Most of the difference in power, or performance per watt, between a laptop GPU and a desktop GPU (or a laptop GPU in an adapter), once both are tuned for efficiency, will come down to the fact that the desktop GPU is placed further away and has it's own PCB and entirely separate power delivery components. Normally I'd consider these losses negligible, but once one is down to the 50-100w range, even a few watts can be significant, which is one of the main reasons laptops have an edge.
If you have performance needs beyond what an APU can deliver and a specific power budget in mind you can estimate the optimal desktop GPU for it. My general rule of thumb is that you can cut the stock TDP of a desktop card in half before a lower-end model (of the same architecture) would be appreciably more efficient.
Last edited: