News The Zen 5 Gaming postmortem: Larger generational gains than many reported, game-boosting Windows Update tested, Ryzen 5 7600X3D gaming benchmarks, too

[thestryker]

Yeah I dropped my sub a while back from HUB... and J2C. I've seen other channels like GN say the 9000 series isn't very good.

I dunno man... I went for it because the 7950X was $523 and the 9950X was $623. For $100 I went with Zen 5 and the added performance even if it is only 10-15%.

What's funny though is I just googled Cinebench R23 scores and found this: R23 scores

It shows a 9950X at 41k for multi-core. I ran the test the other night on my system and scored 43,323 in multicore. I'm at work or I'd post a screenshot.

Point being... the 9000 series isn't even close to being "bad."

It's a bad value, your experience even backs this up. This doesn't mean it isn't worth buying depending on your usage.

Nobody reputable has said that the product by itself is bad just that where it sits in the market is. This is a fair angle to take, but it requires people understanding where they're coming from. Having looked through the reviews of all 4 parts personally speaking the 9950X is the only one worth buying and that's due to its comparative efficiency gains. Down the road once prices stabilize they'll all be the AMD chips to buy for their respective performance.

All that being said the 9000 series are not selling well on the retail market. It seems like people have spoken with their wallets and aren't impressed with the performance per dollar.

 

[TeamRed2024]

Having looked through the reviews of all 4 parts personally speaking the 9950X is the only one worth buying and that's due to its comparative efficiency gains. Down the road once prices stabilize they'll all be the AMD chips to buy for their respective performance.

All that being said the 9000 series are not selling well on the retail market. It seems like people have spoken with their wallets and aren't impressed with the performance per dollar.

Yeah I've read that they aren't selling well... the price dropped from $649 to $623 when I bought it. As for performance per dollar I just said to myself "10% for $100 over the 7950X... that's $10 per 1%."

Not a bad performance (increase) per dollar to me... but others may think differently. The 9950X was the one I considered and agree it's the only one worth buying. Had it been $200-$300 more I'd have went 7950X.

 

[stuff and nonesense]

All that being said the 9000 series are not selling well on the retail market. It seems like people have spoken with their wallets and aren't impressed with the performance per dollar.

Or gamers are waiting for x3d…. Others waiting for the price to drop a little (or a lot).
The 9900x was cheaper numerically and cheaper further with inflation taken into account than the 3900x it replaced was at purchase.. I’m happy with my choice..

 

[thestryker]

Or gamers are waiting for x3d…. Others waiting for the price to drop a little (or a lot).

Who knows the exact reasons, but for example the 7800X3D has gone up in price and its availability has been worse indicates people were waiting and not impressed. The sales figures are also significantly worse than 7000 was at launch and their overall sales rankings are low. All of this points towards people just not seeing the value in them.

The 9900x was cheaper numerically and cheaper further with inflation taken into account than the 3900x it replaced was at purchase..

You could have gotten a 7900X for about 25% less money or a 7950X for about 10% more. This is the value aspect everyone has been speaking of.

I’m happy with my choice..

I'm glad you are, and it's good that you are as that's really all that matters. If it's good for you and your workload that's quite literally the only metric there needs to be.

 

[bit_user]

Having looked through the reviews of all 4 parts personally speaking the 9950X is the only one worth buying and that's due to its comparative efficiency gains. Down the road once prices stabilize they'll all be the AMD chips to buy for their respective performance.

If I were buying an 8-core AM5 CPU today, I'd pay the extra to go with the 9700X. That's mainly because I care more about performance and efficiency than cost.

The 9000 series seems like it was released prematurely, but it appears to be maturing well. For instance, this just dropped, today:

https://www.tomshardware.com/pc-com...ency-cut-in-half-to-match-previous-gen-models
​

I'm really looking forward to updated Ryzen 9000 benchmarks vs. Arrow Lake. I think some people are going to be surprised by how much the Ryzen 9000 models will have improved, at least compared with their Ryzen 7000 counterparts.

 

[TeamRed2024]

The 9000 series seems like it was released prematurely, but it appears to be maturing well. For instance, this just dropped, today:

​

https://www.tomshardware.com/pc-com...ency-cut-in-half-to-match-previous-gen-models​

​

I'm really looking forward to updated Ryzen 9000 benchmarks vs. Arrow Lake. I think some people are going to be surprised by how much the Ryzen 9000 models will have improved, at least compared with their Ryzen 7000 counterparts.

Thanks for that info! Will update my benchmark results once I get the update from MSI.

 

[TheHerald]

Having looked through the reviews of all 4 parts personally speaking the 9950X is the only one worth buying and that's due to its comparative efficiency gains.

Which is the exact same thing with last gen, at the time of release only the 7950x made sense. The rest were overpriced low core options, hence didn't sell well.

 

[larsen0711]

What am I missing? The chart shows a 6.5% performance increase between 7700x and 9700x. How is this larger than others have reported? Looks very similar to TPU results. Anders - https://www.greenlaptops.dk/

 

[stuff and nonesense]

Who knows the exact reasons, but for example the 7800X3D has gone up in price and its availability has been worse indicates people were waiting and not impressed. The sales figures are also significantly worse than 7000 was at launch and their overall sales rankings are low. All of this points towards people just not seeing the value in them.

I’d suggest the following:

5000 series to 7000 series was a relatively large jump in performance. 7000 to 9000 isn’t so great, there is little reason to upgrade from 7000 if you have a cpu from that generation.
The bundle offers on 7000 chips are attractive, I’d guess AMD hopes to flush the supply hence they are being offered at good prices.

For me the 7000 series is a V1 product with the change to DDR 5, so I held off until the major bugs were ironed out..V2 presented its own idiosyncrasies as documented across the press…

 

[thestryker]

For me the 7000 series is a V1 product with the change to DDR 5, so I held off until the major bugs were ironed out..V2 presented its own idiosyncrasies as documented across the press…

Honestly after watching the way things went with AM4 I mostly decided that AMD is always going to have weird little issues so I wasn't too worried about that though I totally understand where you're coming from.

The reason I ended up not going with the 7000 series was actually the IHS being junk. Intel used too much power for my taste across the breadth of workloads I use my system for, but AMD hurt their own efficiency and ran hotter than necessary just to maintain AM4 compatibility. I had hoped this would change with the 9000 series, but it hasn't as Zen 5 is just a more performant core running in a lower power envelope. AMD also didn't change the IOD between Zen 4 and 5 which isn't necessarily a deal breaker, but it does mean platform wise nothing is going to change between the two series.

I still really need to update my system, but I'm not sold on Zen 5 as it stands now so it's really easy for me to wait and see what Intel delivers. It's not impossible that I'll end up punting until next year, but I've been planning an upgrade since 2021 so at some point I just need to do it.

 

[Hotrod2go]

Honestly after watching the way things went with AM4 I mostly decided that AMD is always going to have weird little issues so I wasn't too worried about that though I totally understand where you're coming from.

The reason I ended up not going with the 7000 series was actually the IHS being junk. Intel used too much power for my taste across the breadth of workloads I use my system for, but AMD hurt their own efficiency and ran hotter than necessary just to maintain AM4 compatibility. I had hoped this would change with the 9000 series, but it hasn't as Zen 5 is just a more performant core running in a lower power envelope. AMD also didn't change the IOD between Zen 4 and 5 which isn't necessarily a deal breaker, but it does mean platform wise nothing is going to change between the two series.

I still really need to update my system, but I'm not sold on Zen 5 as it stands now so it's really easy for me to wait and see what Intel delivers. It's not impossible that I'll end up punting until next year, but I've been planning an upgrade since 2021 so at some point I just need to do it.

Basically, depends how the cpu is cooled. Remember AMD moved the thermal sensors in Zen 5 compared to Zen 4 chips. So its difficult to compare temps between the 2 generations unless all factors being equal.
I have found my 9700X a lot easier to cool with the right equipment & that's not going overboard with elaborate water setups. I can't recommend enough the Thermaltake toughair 710 tower cooler for the job, best air cooling device I've ever used in many years. Combine it with Thermal Grizzley Kyronaut extreme TIM & your on a winner.

 

[Jagar123]

I still really need to update my system, but I'm not sold on Zen 5 as it stands now so it's really easy for me to wait and see what Intel delivers. It's not impossible that I'll end up punting until next year, but I've been planning an upgrade since 2021 so at some point I just need to do it.

Yeah, I am in the same boat. I've been wanting to upgrade for a couple of years. I typically upgrade both my CPU and GPU at the same time, but seeing as GPUs have been stupid lately I think I'll just need to upgrade my CPU for now and wait on the GPU upgrade. If the 7800X3D was at the $340 price it had been at I would likely buy it today and get it done. Sadly, it has spiked back up to near MSRP so I'll continue to wait...

 

[thestryker]

Basically, depends how the cpu is cooled. Remember AMD moved the thermal sensors in Zen 5 compared to Zen 4 chips. So its difficult to compare temps between the 2 generations unless all factors being equal.
I have found my 9700X a lot easier to cool with the right equipment & that's not going overboard with elaborate water setups. I can't recommend enough the Thermaltake toughair 710 tower cooler for the job, best air cooling device I've ever used in many years. Combine it with Thermal Grizzley Kyronaut extreme TIM & your on a winner.

Changing how they've handled the thermal sensors has nothing to do with the bad IHS and it impacting efficiency due to poor heat transfer. The die sizes are virtually identical, and the IOD is identical, so there are minimal improvements they could make at the die level which could impact this issue.

 

[Eximo]

Changing how they've handled the thermal sensors has nothing to do with the bad IHS and it impacting efficiency due to poor heat transfer. The die sizes are virtually identical, and the IOD is identical, so there are minimal improvements they could make at the die level which could impact this issue.

Could always go direct die and/or liquid metal if you are up for it.

I guess the question comes down to how much in matters in terms of performance to you to have the die be a little cooler.

I like to water cool, mostly because it is fun to build it, the results are fine lately. So I haven't felt the need to de-lid since my 7700k and its one really warm core.

 

[thestryker]

Could always go direct die and/or liquid metal if you are up for it.

I guess the question comes down to how much in matters in terms of performance to you to have the die be a little cooler.

I've opted to not buy at all rather than support their bad design and void the warranty to fix it. I wouldn't buy anything from the 7000 or 9000 series without a long term plan as far as direct die/replacing the IHS is concerned (I'd only be looking at 16 core) though. The benefit of the lower power parts (and X3D to an extent) is that they don't use enough power for the bad IHS to really be a problem.

I like to water cool, mostly because it is fun to build it, the results are fine lately. So I haven't felt the need to de-lid since my 7700k and its one really warm core.

When I installed my first AIO and got video card temperature reductions because of moving the heat away from the CPU socket alone made me want to always use one going forward. It's certainly possible to get passable cooling, but AMD just not designing a bad IHS would be better.

 

[bit_user]

I've opted to not buy at all rather than support their bad design and void the warranty to fix it. I wouldn't buy anything from the 7000 or 9000 series without a long term plan as far as direct die/replacing the IHS is concerned (I'd only be looking at 16 core) though. The benefit of the lower power parts (and X3D to an extent) is that they don't use enough power for the bad IHS to really be a problem.

You can achieve a similar result by putting it in ECO mode or otherwise reducing the power limits.

It's certainly possible to get passable cooling, but AMD just not designing a bad IHS would be better.

Just because it's a bit thick?

I think you're being too hard on AMD, here. A big part of the thermal bottlenecking problem, in the AM5 CPUs, is due to their greater thermal density. Do the math, yourself. Divide Raptor Lake's PL2 by its die area and compare that with the PPT of the 7950 (minus whatever you estimate the IOD is using), divided by their area. It's much higher.

Intel likely ran up against the same problem with Meteor Lake-S and Arrow Lake. Why do you think Arrow Lake doesn't clock as high as Raptor Lake did?

 

[thestryker]

You can achieve a similar result by putting it in ECO mode or otherwise reducing the power limits.

Yeah and lose performance in the process which shouldn't be necessary to fix a problem that shouldn't exist in the first place.

Just because it's a bit thick?

I think you're being too hard on AMD, here.

Really? When you can remove the IHS surface down to a reasonable amount and see real world power consumption and temperature benefits I don't really think I am.

 

[bit_user]

Really? When you can remove the IHS surface down to a reasonable amount and see real world power consumption and temperature benefits I don't really think I am.

Where have you seen data on someone merely thinning the AM5 IHS?

 

[thestryker]

Where have you seen data on someone merely thinning the AM5 IHS?

I hate trying to find YT videos so I was hoping you wouldn't ask. I looked for Roman's video that he did on it as well but didn't see it and I wasn't going through all of the videos of that rough time period to try to find it.

View: https://www.youtube.com/watch?v=vmQ7IU8Nj2c


The TG tool for it with detailed explanation: https://www.thermal-grizzly.com/en/lapping-tool/s-tg-lt-ar7000

 

[bit_user]

I hate trying to find YT videos so I was hoping you wouldn't ask. I looked for Roman's video that he did on it as well but didn't see it and I wasn't going through all of the videos of that rough time period to try to find it.

View: https://www.youtube.com/watch?v=vmQ7IU8Nj2c


The TG tool for it with detailed explanation: https://www.thermal-grizzly.com/en/lapping-tool/s-tg-lt-ar7000

Okay, so the web page link tells us the guy in the video achieved a 5 C reduction on a 7950X with a 360 mm AIO. All of the other examples they mentioned seem to involve changing more than one variable.

With a less capable cooler or a less power-hungry model, the IHS should be less of a bottlenecks.

The other thing is, we know that even greater gains are possible with direct-die water cooling. So, in a way, thinning the IHS is just a half-step towards that. I'll bet you'd see a similar effect from thinning Raptor Lake's IHS.

Also, the page you linked points out that lapping the IHS evens out any curvature or warping and occurred during manufacturing. I wonder how much of the drop is just from that. I definitely recall people talking about gains on that order, just by lapping the IHS and heatsink to achieve a flatter mating surface, not even trying to remove a significant amount of material from the IHS.

 

[TheHerald]

Where have you seen data on someone merely thinning the AM5 IHS?

There are aftermarket ihs for am5, the temp reductions are 10+c on the big chips.

 

[thestryker]

I'll bet you'd see a similar effect from thinning Raptor Lake's IHS.

Highly unlikely since there isn't as much to remove as they note the typical limit is 0.2mm and with AM5 it's 1mm. The IHS is comparatively absurdly thick which definitely has an impact on cooling. Roman did some metallurgical tests because people had posited maybe the AMD IHS was lower quality, but it was basically identical to Intel's. He also did a video which I think was before the Zen 4 launch talking about why a thick IHS is bad.

Also, the page you linked points out that lapping the IHS evens out any curvature or warping and occurred during manufacturing. I wonder how much of the drop is just from that. I definitely recall people talking about gains on that order, just by lapping the IHS and heatsink to achieve a flatter mating surface, not even trying to remove a significant amount of material from the IHS.

The benefit of a flatter surface depends mostly on the cooler. Most current coolers take into account IHS curvature and aren't flat which is one of the reasons lapping has fallen out of favor.

With a less capable cooler or a less power-hungry model, the IHS should be less of a bottlenecks.

Yeah the X3D and 65W CPUs shouldn't really see much of an impact because they're generally going to be power limited or clock limited. The efficiency gains are low enough to not be meaningful for the low power usage parts.

 

[-Fran-]

Okay, so the web page link tells us the guy in the video achieved a 5 C reduction on a 7950X with a 360 mm AIO. All of the other examples they mentioned seem to involve changing more than one variable.

With a less capable cooler or a less power-hungry model, the IHS should be less of a bottlenecks.

The other thing is, we know that even greater gains are possible with direct-die water cooling. So, in a way, thinning the IHS is just a half-step towards that. I'll bet you'd see a similar effect from thinning Raptor Lake's IHS.

Also, the page you linked points out that lapping the IHS evens out any curvature or warping and occurred during manufacturing. I wonder how much of the drop is just from that. I definitely recall people talking about gains on that order, just by lapping the IHS and heatsink to achieve a flatter mating surface, not even trying to remove a significant amount of material from the IHS.

The IHS is bad. That was proven on day1 release for Zen4, which is sad for AMD. They did one thing right and one thing wrong. Too bad the good wasn't as good to offset the bad.

The good is keeping backwards compatibility with coolers. The bad is that, on average, it's 15°C detriment for the CPU against delidded under the same cooling conditions. The one who investigated this extensively was der8auer. This is his initial video on the topic:

View: https://www.youtube.com/watch?v=y_jaS_FZcjI


And a related one for temp scaling:

View: https://www.youtube.com/watch?v=0mPi6DjGZpQ


That one accounts for the "temperature condensation" of the dies and their place on the chip interposer thing.

In any case, I was hoping AMD would be improving the IHS itself, but if they're to be believed, their problem was more with the way they were getting the temps and placement and not the "real" temps, so they changed that with Zen5 and seems ok so far; I haven't seen or read any reports on weirdness due to this change. Then again, no one is buying them, so very little data

Regards.

 

[bit_user]

The bad is that, on average, it's 15°C detriment for the CPU against delidded under the same cooling conditions.

This is a bogus comparison. Direct-die water cooling has always yielded like low double-digit improvements on just about any CPU. People are even direct-die cooling Raptor Lake!

 

[bit_user]

Highly unlikely since there isn't as much to remove as they note the typical limit is 0.2mm and with AM5 it's 1mm. The IHS is comparatively absurdly thick which definitely has an impact on cooling. Roman did some metallurgical tests because people had posited maybe the AMD IHS was lower quality, but it was basically identical to Intel's. He also did a video which I think was before the Zen 4 launch talking about why a thick IHS is bad.

The benefit of a flatter surface depends mostly on the cooler. Most current coolers take into account IHS curvature and aren't flat which is one of the reasons lapping has fallen out of favor.

Well, that's interesting because I found this:

"on a lapped CPU with a lapped cooler, der8auer was able to reduce his Intel CPUs' junction temperatures by as much as 7.1ºC."

https://hothardware.com/news/alder-lake-ilm-mod-big-drop-in-core-i9-12900k-cpu-temps
​