RTX 4060 debate, potentially not terrible? *Nope, it is.

[WallacEngineering]

7 minutes ago, Agall said:

I'd argue that the A380 offers the lowest price for a dedicated AV1 encoder and quad display outputs, being a product rich with features that the competition doesn't have at that price point.

 

Ya but look at where its performance is. This is pretty much why nobody buys them.

 

Ive heard about Intel's upcoming Battlemage GPUs and I have high hopes they will be significantly better. They still won't match AMD or Nvidia, but I have a feeling they won't have a problem matching Mid-Tier GPUs, kind of like AMD's 5700-XT did.

 

[Agall]

4 minutes ago, WallacEngineering said:

 

Ya but look at where its performance is. This is pretty much why nobody buys them.

 

Ive heard about Intel's upcoming Battlemage GPUs and I have high hopes they will be significantly better. They still won't match AMD or Nvidia, but I have a feeling they won't have a problem matching Mid-Tier GPUs, kind of like AMD's 5700-XT did.

 

Still plenty of people with a GTX 1060 6GB, which the A380 is getting ever closer to in performance.

[WallacEngineering]

27 minutes ago, Agall said:

Still plenty of people with a GTX 1060 6GB, which the A380 is getting ever closer to in performance.

 

Ya but 1060 is basically dead at this point and people are looking to upgrade from it.

 

Only the 1080 and 1080-Ti really have any life left in them, maybe 1070 just because it at least has 8GB of VRAM.

 

Whats really surprising to me is just how horrible the RX 6500 and RTX 3050 are compared to the rest of their product stacks - a real lession in why you should really stay away from the very lowest-tier GPUs. At least the RTX 3050 isn't completely terrible, but I certainly wouldn't buy one.

 

If I wanted a lower-tier Nvidia card right now I would just get an RTX 3060 12GB and call it a day. But of course since Im fine with AMD it would definitely be the RX 6700-XT without a doubt.

 

EDIT: And holy crap I didn't even see the GTX 1650 because its literally down at the very bottom edge of the chart. How is it that bad? This chart is taken from an average of 10 games, so how did it completely fail at every single one of them???

[Agall]

25 minutes ago, WallacEngineering said:

 

Ya but 1060 is basically dead at this point and people are looking to upgrade from it.

 

Only the 1080 and 1080-Ti really have any life left in them, maybe 1070 just because it at least has 8GB of VRAM.

 

Whats really surprising to me is just how horrible the RX 6500 and RTX 3050 are compared to the rest of their product stacks - a real lession in why you should really stay away from the very lowest-tier GPUs.

 

EDIT: And holy crap I didn't even see the GTX 1650 because its literally down at the very bottom edge of the chart. How is it that bad? This chart is taken from an average of 10 games, so how did it completely fail at every single one of them???

I consider a GTX 1070ti and greater still viable today for most users. The argument however is different when you're talking $120 compared to something like the RX 6500 XT at $160. Especially considering it doesn't have quad outputs and a dedicated encoder, so its always useful in some context outside of gaming.

 

Low tier cards will always have a poor argument with regards to price/performance, since it doesn't scale well at its price, but at least the A380 doesn't suffer from the other disadvantages of low tier cards that only come with two display outputs and no dedicated encoder while also having less VRAM. 

 

I'll be able to find the limitations of an A380 here soon, 12100f on the way to support it as a sub $500 new gaming PC. A more reasonable choice would obviously be to double the price of the GPU and opt for an RX 6600, but that's not the point.

 

Technically a 3 dimensional graph, but I broke it down to two dimensions, but this is how I view this discussion. Yes the A380 falls into 'low end' but minimum price purchases on products are sometimes done out of necessity, which an A380 is drastically better than an iGPU.

 

 

 

 

 

[WallacEngineering]

3 minutes ago, Agall said:

I consider a GTX 1070ti and greater still viable today for most users. The argument however is different when you're talking $120 compared to something like the RX 6500 XT at $160. Especially considering it doesn't have quad outputs and a dedicated encoder, so its always useful in some context outside of gaming.

 

Low tier cards will always have a poor argument with regards to price/performance, since it doesn't scale well at its price, but at least the A380 doesn't suffer from the other disadvantages of low tier cards that only come with two display outputs and no dedicated encoder while also having less VRAM. 

 

I'll be able to find the limitations of an A380 here soon, 12100f on the way to support it as a sub $500 new gaming PC. A more reasonable choice would obviously be to double the price of the GPU and opt for an RX 6600, but that's not the point.

 

 

 

Will be interesting to see what the little A380 can do, I wish you all the best in your experiments

[Agall]

4 minutes ago, WallacEngineering said:

 

Will be interesting to see what the little A380 can do, I wish you all the best in your experiments

Slightly adjusted the graphical representation.

 

If it puts it into perspective, the hardware isn't going to end up as a gaming PC but for another purpose, I'll just be testing it as a gaming PC. A380 as well is one of the best test GPU for any of us that might need one for a friend/family's system, especially in the era of iGPU-less CPUs.

[r00tb33r]

A380 plays Quake 2 RTX at a glorious resolution of 960x600 (16:10 monitor), maintaining 60fps.  Anymore than that and it will have to dip or scale down dynamically.

 

Neat that it has the capability, but too little power to be useful.

[Agall]

2 minutes ago, r00tb33r said:

A380 plays Quake 2 RTX at a glorious resolution of 960x600 (16:10 monitor), maintaining 60fps.  Anymore than that and it will have to dip or scale down dynamically.

 

Neat that it has the capability, but too little power to be useful.

What's better value? An A380 or RTX 4060?

[r00tb33r]

11 minutes ago, Agall said:

What's better value? An A380 or RTX 4060?

I don't have a 4060 but I am trying to offload my A380 since I no longer need it for work.  It's worth a lot less now, to the point it might be not worth selling it, but it is also not useful to me anymore.

 

Amusingly my DG1 cards are a lot more useful, even though it would be even harder to sell them.

 

A 4060 would have been useful.  I just bought a 12GB 3060 for peanuts though, so I'm probably good for a while.

[Agall]

1 minute ago, r00tb33r said:

I don't have a 4060 but I am trying to offload my A380 since I no longer need it for work.  It's worth a lot less now, to the point it might be not worth selling it, but it is also not useful to me anymore.

 

Amusingly my DG1 cards are a lot more useful, even though it would be even harder to sell them.

I don't believe that answers the question of which you'd consider a better value, an A380 at $120 or RTX 4060 at $300?

[r00tb33r]

3 minutes ago, Agall said:

I don't believe that answers the question of which you'd consider a better value, an A380 at $120 or RTX 4060 at $300?

4060 is a better value.  But I'd just get 30 series instead if you're looking for value.

 

I edited the earlier post.

[WallacEngineering]

3 hours ago, r00tb33r said:

4060 is a better value.  But I'd just get 30 series instead if you're looking for value.

 

I edited the earlier post.

 

Yep,

 

Used RTX 3060 12GB if you want low-mid range Nvidia for a good value.

[r00tb33r]

 

 

So gen3 doesn't make much of a difference, which is a pleasant surprise.  Price aside this is starting to look like a good card.

 

[WallacEngineering]

30 minutes ago, r00tb33r said:

So gen3 doesn't make much of a difference, which is a pleasant surprise.  Price aside this is starting to look like a good card.

 

 

Um, what? We just spent the better part of a few days all saying it is garbage.

 

Of course when you compare it to a 2-generations old card, its going to look pretty good lol

[r00tb33r]

6 minutes ago, WallacEngineering said:

Um, what? We just spent the better part of a few days all saying it is garbage.

Sorry if your effort is a waste.

[WallacEngineering]

12 minutes ago, r00tb33r said:

Sorry if your effort is a waste.

Well, ur the only one who likes it so far

[r00tb33r]

3 minutes ago, WallacEngineering said:

Well, ur the only one who likes it so far

Someone has to buy it first so I could score a used deal.

[r00tb33r]

 

[bad-3_j]

ONe small video explains it all.

 

 

[RevGAM]

8 hours ago, r00tb33r said:

 

 

I have mixed feelings about what he said. He's correct but I feel like he missed the bull's-eye. Anyone else?

[WallacEngineering]

10 hours ago, r00tb33r said:

 

 

@RevGAM

@Agall

 

Theres a few things that are right here, and a few things that are wrong. So whats Wrong:

 

Well, testing on a high-end or low end system doesn't matter, so long as you turn the game settings up high enough to stress the card and not the rest of the system.

 

That RTX 4060 will perform exactly the same on that Ryzen 5600 as it would on an Intel i9-13900K system, so long as the GPU is sitting at 100% utilization, this has already been proven time and time again. So yes, testing with a top-of-the-line system is actually the CORRECT way to do it so you compare the graphics cards to eachother, and NOT the rest of the system.

 

If you review a GPU using a system that bottlenecks your GPU even by 0.10% by using a CPU that cannot process finished frames to the monitor fast enough to absolutely maximize performance, then your entire testing methodology is rendered completely invalid and useless to anyone who didn't buy the EXACT same system, which means the review is pointless to everyone else.

 

However, if your test system is so powerful that there is no way whatsoever for someone to buy a system that limits the performance of the graphics card, then your review IS valid and applicable to anyone buying that same graphics card, because they will see the same performance so long as they actually use the card to its fullest extent.

 

Secondly, DLSS and FSR should NEVER be included in performance comparisons. Fake Frame Generation and Upscaling DO NOT COUNT, no matter how much performance it may give you. Your girlfriend can fake her orgasm too, but would you really consider that a "Feature"? No, I didn't think so. Its cool having DLSS-3, sure, but it is NOT a reason to spend money on a new card.

 

But one thing I 100% agree with - especially with ever-increasing VRAM demand in games - is YouTubers should include testing at High Settings, not just Ultra - also because Ultra is such a high performance hit for not much fidelity increase and so nobody but owners of Top-Tier flagship cards ever run Ultra Settings on drmanding titles.

 

Also, his moving around of the product stack at the end of the video helps to make the series make more sense, but he also moved the price tiers down a slot and that just isn't the reality. If you could pay $699 or whatever for RTX 4080 performance, then I don't think anyone would have any complaints. But its not $700, its $1200. Also moving the product stack like that would mean that previous gen RTX 3000 cards are actually faster than their new RTX 4000 replacements in some cases, and that would be an even WORSE disaster.

 

But he is certainly correct about companies trying to upsell you not necessarily being the worst idea, and his reference to a car dealership is pretty much on point.

[Agall]

10 minutes ago, WallacEngineering said:

@RevGAM

@Agall

 

Theres a few things that are right here, and a few things that are wrong. So whats Wrong:

 

Well, testing on a high-end or low end system doesn't matter, so long as you turn the game settings up high enough to stress the card and not the rest of the system.

 

That RTX 4060 will perform exactly the same on that Ryzen 5600 as it would on an Intel i9-13900K system, so long as the GPU is sitting at 100% utilization, this has already been proven time and time again. So yes, testing with a top-of-the-line system is actually the CORRECT way to do it so you compare the graphics cards to eachother, and NOT the rest of the system.

 

Secondly, DLSS and FSR should NEVER be included in performance comparisons. Fake Frame Generation and Upscaling DO NOT COUNT, no matter how much performance it may give you. Your girlfriend can fake her orgasm too, but would you really consider that a "Feature"? No, I didn't think so. Its cool having DLSS-3, sure, but it is NOT a reason to spend money on a new card.

 

But one thing I 100% agree with - especially with ever-increasing VRAM demand in games - is YouTubers should include testing at High Settings, not just Ultra - also because Ultra is such a high performance hit for not much fidelity increase.

The 5600 and 13900k will not perform the same depending on the specific application, whether its an Arc A380 or 4090. Its the whole discussion I have about MMOs/multiplayer games.

 

Plenty of GPUs can exceed 60 fps in almost every scenario but will regularly drop below unless you're running the newest IPC/frequency/cache monster of a CPU. Its a limitation that's always been there but has only been noticeable with the substantial gap in CPU performance we've had in the last two generations.

[WallacEngineering]

34 minutes ago, Agall said:

The 5600 and 13900k will not perform the same depending on the specific application, whether its an Arc A380 or 4090. Its the whole discussion I have about MMOs/multiplayer games.

 

Plenty of GPUs can exceed 60 fps in almost every scenario but will regularly drop below unless you're running the newest IPC/frequency/cache monster of a CPU. Its a limitation that's always been there but has only been noticeable with the substantial gap in CPU performance we've had in the last two generations.

 

Yes but Im talking about the testing methodology being correct because A.) There isn't another option and building 10 different systems to run 100 tests 10-times over to test one GPU using varying tiers of CPUs is simply out of the question and 2.) You turn graphical settings up high enough to be bottlenecked on the GPU, not the CPU, in which case it doesn't matter if the CPU is vastly more powerful.

 

I mean it might matter if you run it against a 10-year old CPU, sure, but as long as the CPU is modern and of a decent tier, an RTX 4060 will run at 30 FPS at 4K High whether the CPU is a Ryzen 5600 or i9-13900K. There might be a hint of a difference, maybe 1% or so, but the GPU is the part of the PC that is overloaded - and thats the point of GPU-specific reviews. Then you do the opposite - Run an RTX 4090 at 1080p low-medium settings to skyrocket FPS and show the bottleneck of CPUs in CPU-specific reviews. So there is really no other way to do testing and thats what that YouTuber just doesn't understand.

[Sir Beregond]

On 6/28/2023 at 8:13 AM, WallacEngineering said:

 

Ya and like Ive literally told you a thousand times now, I don't care about the classes or names either - only the price to performance ratio, that you for some reason still can't get through your head

Names and classes are fluid every gen and mean next to nothing. 

 

Always look at the silicon.

[WallacEngineering]

6 minutes ago, Sir Beregond said:

Names and classes are fluid every gen and mean next to nothing. 

 

Always look at the silicon.

 

Thats exactly what I was doing, read all the comments before commenting on one