GPUs & CPUs & Enthusiast hardware: Questions, Discussion and fanboy slap-fights - Nvidia & AMD & Intel - Separe but Equal. Intel rides in the back of the bus.

[Basil II]

When GPU spergery goes too far...

 

[Dante Alighieri]

When GPU spergery goes too far...

Of course the killer is a nvidia fan

 

[The Mass Shooter Ron Soye]

Leaked slide: https://www.technopat.net/wp-content/uploads/2020/10/Ryzen-5000-DDR4-4000.jpg

lmao AMD went full jew, surely this won't have any negative effects for this company.

When your product is better than the competition in every way, you set the price.

Read that the Ryzen 5600 will retail at $220 and be released in early 2021. That's just 20 bananas up from the previous R5 x600s and 80 bucks cheaper than the 5600X. The new X versions seems to have that early adopter tax.

5600X price could decline by the time the 5600 comes out, and definitely by Intel's Rocket Lake launch, even if Rocket Lake only "ties" Zen 3.

It will be interesting to see what they put in between the 5600X and 5800X. The 5800X is the worst priced chip of the lineup and the 5600X at least comes with a cooler. There's also a $150 gap in between 5600X and 5800X right now.

 

[mr.moon1488]

I don't really care much about the whole Nvidia V. AMD debate, but I'm most likely going to get a Nvidia because I already have GeForce Experience installed and don't want to go through the trouble of installing whatever AMD uses to do updates.

 

[CreamyHerman’s]

I don't really care much about the whole Nvidia V. AMD debate, but I'm most likely going to get a Nvidia because I already have GeForce Experience installed and don't want to go through the trouble of installing whatever AMD uses to do updates.

AMD updates are surprisingly hassle free and you don’t need a email

 

[Smaug's Smokey Hole]

AMD doesn't have the option to install simple GPU drivers like Nvidia where you just get drivers and the old Nvidia Control Panel. On the other hand the Radeon software is not even close to the annoyance that GeForce Experience is.

Leaked slide: https://www.technopat.net/wp-content/uploads/2020/10/Ryzen-5000-DDR4-4000.jpg

If everything in that slide is correct then they're suggesting that Infinity Fabric can run at 2000mhz on every CPU out of the box, no problem. That's quite the leap and I think that on Ryzen 3X00 IF started crapping out if overclocked above 1800mhz(up from the default 1600mhz). For those not familiar with this nerdery it means buying super fast RAM might be worth it.

 

[ZMOT]

AMD doesn't have the option to install simple GPU drivers like Nvidia where you just get drivers and the old Nvidia Control Panel. On the other hand the Radeon software is not even close to the annoyance that GeForce Experience is.

another difference is with amd you don't have to install the newest driver every month for nvidia™ geforce™ RTX™ optimized™ features™ (amd doesn't have the money or manpower to inject people into devteams or optimize every game), so shit will be a few frames slower but generally work. on amd never had to backdate a driver because shit outright didn't work with the newest one or the newest release absolute shite (inb4 amd driver memes)

 

[Basil II]

another difference is with amd you don't have to install the newest driver every month for nvidia™ geforce™ RTX™ optimized™ features™ (amd doesn't have the money or manpower to inject people into devteams or optimize every game), so shit will be a few frames slower but generally work. on amd never had to backdate a driver because shit outright didn't work with the newest one or the newest release absolute shite (inb4 amd driver memes)

also AMD has better drivers for nonfisher-price Operating systems like Linux.

 

[Stasi]

I just got a small form factor machine with a Ryzen 3 2200 apu to use as a living room PC for multi media and light gaming.

Are there any settings I should mess about with, tweaks etc with the apu? Tested with a couple of games and seems to work OK but I don't know if there is anything specific I need to do with an apu as opposed to traditional CPU/GPU combo to get the best out of it.

 

[Vecr]

I just got a small form factor machine with a Ryzen 3 2200 apu to use as a living room PC for multi media and light gaming.

Are there any settings I should mess about with, tweaks etc with the apu? Tested with a couple of games and seems to work OK but I don't know if there is anything specific I need to do with an apu as opposed to traditional CPU/GPU combo to get the best out of it.

I guess you could try overclocking the RAM, depending on the APU that is used for most or all of the video RAM, instead of it being on the video card itself. Unless you are having issues, it's probably not needed.

 

[CreamyHerman’s]

I just got a small form factor machine with a Ryzen 3 2200 apu to use as a living room PC for multi media and light gaming.

Are there any settings I should mess about with, tweaks etc with the apu? Tested with a couple of games and seems to work OK but I don't know if there is anything specific I need to do with an apu as opposed to traditional CPU/GPU combo to get the best out of it.

Are you using Dual Channel? You can increase your FPS by 30% using Dual Channel

 

[Smaug's Smokey Hole]

Are you using Dual Channel? You can increase your FPS by 30% using Dual Channel

Good point, if it's pre-built it might ship with one stick and one unused slot, add another one in and performance goes vroom.

Also, give Ryzen Master a try if you want to try overclocking, a free +200mhz without changing anything should be possible, at worst windows hangs and you have to restart. There's no risk, components don't burn themselves out that way anymore.

 

[CreamyHerman’s]

Good point, if it's pre-built it might ship with one stick and one unused slot, add another one in and performance goes vroom.

Also, give Ryzen Master a try if you want to try overclocking, a free +200mhz without changing anything should be possible, at worst windows hangs and you have to restart. There's no risk, components don't burn themselves out that way anymore.

There is a good reason why this is: Memory Bandwidth for the APU. One stick is about 64-bit bandwidth, two is double that.

 

[Stasi]

There is a good reason why this is: Memory Bandwidth for the APU. One stick is about 64-bit bandwidth, two is double that.

Thanks for the tip, its a pre-built business machine so like @Smaug's Smokey Hole said it comes with one 8gb stick. I'll add another stick of RAM to my shopping list.

 

[Smaug's Smokey Hole]

Thanks for the tip, its a pre-built business machine so like @Smaug's Smokey Hole said it comes with one 8gb stick. I'll add another stick of RAM to my shopping list.

Check the RAM speed as well, it's probably DDR4 3200 memory but it might be set to 2400. Download CPU-Z and check the memory tab just in case.

 

[Smaug's Smokey Hole]

	RTX 3080	RTX 3070 TI/Super?	RTX 3070
	GA102	GA102	GA104
	8 704 st.	7 424 st.	5 888 st.
	1 710 MHz	?	1 730 MHz
	320-bit	320-bit	256-bit
	10 GB GDDR6X	?	8 GB GDDR6
	19 000 MHz	?	16 000 MHz
	760 GB/s	?	512 GB/s

Some new leaks/speculation around a beefier 3070 to counter AMD. If Nvidia is getting antsy it really looks like AMD have something competitive.

Prediction: The 3070 TI/Super would be 10GB GDDR6 (non-X) and probably between 600-650GB bandwidth, no reason for a big change in core clock and memory frequency will probably be the same.

 

[Just Some Other Guy]

	RTX 3080	RTX 3070 TI/Super?	RTX 3070
	GA102	GA102	GA104
	8 704 st.	7 424 st.	5 888 st.
	1 710 MHz	?	1 730 MHz
	320-bit	320-bit	256-bit
	10 GB GDDR6X	?	8 GB GDDR6
	19 000 MHz	?	16 000 MHz
	760 GB/s	?	512 GB/s

Some new leaks/speculation around a beefier 3070 to counter AMD. If Nvidia is getting antsy it really looks like AMD have something competitive.

Prediction: The 3070 TI/Super would be 10GB GDDR6 (non-X) and probably between 600-650GB bandwidth, no reason for a big change in core clock and memory frequency will probably be the same.

For good reason.

 

[Ginger Piglet]

	RTX 3080	RTX 3070 TI/Super?	RTX 3070
	GA102	GA102	GA104
	8 704 st.	7 424 st.	5 888 st.
	1 710 MHz	?	1 730 MHz
	320-bit	320-bit	256-bit
	10 GB GDDR6X	?	8 GB GDDR6
	19 000 MHz	?	16 000 MHz
	760 GB/s	?	512 GB/s

Some new leaks/speculation around a beefier 3070 to counter AMD. If Nvidia is getting antsy it really looks like AMD have something competitive.

Prediction: The 3070 TI/Super would be 10GB GDDR6 (non-X) and probably between 600-650GB bandwidth, no reason for a big change in core clock and memory frequency will probably be the same.

And also it sounds like AMD are actually going to have cards for sale on launch.

The problem with Ampere is that reportedly, Samsung just can't fabricate the actual chips fast enough. This means that Nvidia, if this is true, have a product that they can't supply in any meaningful volume. In those circumstances, all AMD has to do with Big Navi is have a card available for actual sale that beats the 2080 Ti, even only by a couple percent, and they will clean up. This would also explain why they floated the rumours around a 16/20 GB Ampere, as a form of FUD. If Big Navi is equal to or better than the RTX 3080, Nvidia will be in extreme trouble because that was the exact combination of factors that in 2000 allowed the original Athlon and Athlon Thunderbird to demolish the previously unassailable Pentium III. Intel was stuck making vaporware like the Pentium III 1000 and 1133 and relying on a product which was still only just in development in the Pentium 4 and we all know what a mess that turned out to be.

(I'm still salty to this day about having a shiny new P4 1.5 GHz system back then and showing it off to my mates only for one of them to reveal that his Athlon 1100+ spanked it in every way.)

 

[Smaug's Smokey Hole]

And also it sounds like AMD are actually going to have cards for sale on launch.

The problem with Ampere is that reportedly, Samsung just can't fabricate the actual chips fast enough. This means that Nvidia, if this is true, have a product that they can't supply in any meaningful volume. In those circumstances, all AMD has to do with Big Navi is have a card available for actual sale that beats the 2080 Ti, even only by a couple percent, and they will clean up. This would also explain why they floated the rumours around a 16/20 GB Ampere, as a form of FUD. If Big Navi is equal to or better than the RTX 3080, Nvidia will be in extreme trouble because that was the exact combination of factors that in 2000 allowed the original Athlon and Athlon Thunderbird to demolish the previously unassailable Pentium III. Intel was stuck making vaporware like the Pentium III 1000 and 1133 and relying on a product which was still only just in development in the Pentium 4 and we all know what a mess that turned out to be.

(I'm still salty to this day about having a shiny new P4 1.5 GHz system back then and showing it off to my mates only for one of them to reveal that his Athlon 1100+ spanked it in every way.)

It would be very strange if all of this sort of mirror the Nvidia FX era where ATI(now AMD) were top dog at a great price, it would be strange because the 3000 series doesn't suck like the FX cards did but it being unavailable is like the phantom flagship 5950(I think?) that mostly existed on paper. It was there if you could buy it and you couldn't buy it.

I wonder why they went with Samsung, I think TSMCs process is built/licensed on Samsungs stuff and it works. Personally I've been suspecting GDDR6X manufacturing not being up to snuff but that wouldn't stop them from putting out 3070's unless they're waiting to see what AMD actually have.

 

[Just Some Other Guy]

It would be very strange if all of this sort of mirror the Nvidia FX era where ATI(now AMD) were top dog at a great price, it would be strange because the 3000 series doesn't suck like the FX cards did but it being unavailable is like the phantom flagship 5950(I think?) that mostly existed on paper. It was there if you could buy it and you couldn't buy it.

I wonder why they went with Samsung, I think TSMCs process is built/licensed on Samsungs stuff and it works. Personally I've been suspecting GDDR6X manufacturing not being up to snuff but that wouldn't stop them from putting out 3070's unless they're waiting to see what AMD actually have.

I think they are using 7nm, but only have enough for the professional cards.. Also Samsung apparently gave them a great deal on 8nm, but it's still an inferior silicon and it shows.

Edit - I'm not too surprised if AMD knocks it out of the park. Despite driver issues, the 5700xt showed promising performance. Some forget it was only a 40CU card.

I've actually seen some people wonder how AMD can suddenly have a card 2x (and then some) faster then the last 'flagship'. Easy when they decide to make a full card with a better arch.