AMD Ryzen 5 9600X and Ryzen 7 9700X review roundup : All hail the efficiency kings

[desiibond]

There is no way in this world that anyone (who is not running some kind of mining or similar operation) is saving 400-500 Rs per month by changing just the CPU. These efficiency calculations are for max load. Most times, the CPUs are running close to or under 25W. So, the savings would be less than Rs. 100 per month even for always on computers.

To make it easier to understand, I am bringing laptops into the discussion. AMD Strix Point 28W CPU matches the performance of 50W CPU of previous gen. How does it help?

It is not just direct power savings. Even a 10% increase in performance, with a 30% decrease in power draw brings huge improvement. Take corporate laptops for example. Our laptops are used 9-10 hours a day. We do local code compilation which relies heavily on CPU. My current office laptop is powered by a 12th gen Intel CPU and the battery life sucks donkey balls. If I unplug, I barely get 3-4 hours of battery life. As a result, we all work with laptopsconnected to power most of the time. This laptop is just 1.5 years old and battery degraded to 81%. I now get max of 3 hours of battery life. Reason is highly inefficient chip. By the time the laptop is 2 years old, i will ahve to change the battery and I have no idea how much it costs. At the same time, the performance is average at best. Most of the time CPU is running at 60-80% load. When I initiate a code compilation, it gets hot. When I set it to sleep, the idle battery drain is higher and so I end up keeping it plugged entire night. I do not want to wake up with laptops whose battery is totally drained. Now, imagine how much money we are losing by relying on a CPU that is not power efficient!

Coming to desktops where there is no battery to think of, the savings are not immediately noticed. We use desktops for heavy lifting. Be it gaming or running multiple VMs or containers etc. If I have a CPU that draws less power and performs better, I end up with simulations or rendering or VMs running while the desktop is drawing less power. As drivers get better, the peformance goes up. I am finishing my heavy compilations or rendering faster in cases where max CPU utilized because in these cases, the CPU does not throttle any more as it is not reaching thermal limits. As the job gets done faster, I am saving on power utilization, on component life.

Now, combine these two (desktop + laptop) and you will understand why its not less than 100/- per month.

I have not even touched on servers. Imagine the amount of savings for corporates when a highly efficient CPU line up comes into the company.

A simple example is that ever since I got my MBP powered by M2 Pro, I am doing more work on that one because it consumes so little power and does majority fo the work. This is where AMD is going now, this is where Qualcomm comes into picture. Intel ****ed up PPW in last few years and AMD is bringing this back to front.

 

[AverageSkyrimEnjoyer]

Is it worth waiting for these chips in the Indian market? With a non-inflated price similar to the price of their last gen counterparts.

 

[desiibond]

Is it worth waiting for these chips in the Indian market? With a non-inflated price similar to the price of their last gen counterparts.

Depends on what you want, what your budget is and what kind of urgency you are in. There will be stability issues for sure for few weeks, with 9000 cpus. Do not try to be the early adopter.

 

[Psycho_McCrazy]

I have not even touched on servers. Imagine the amount of savings for corporates when a highly efficient CPU line up comes into the company.

Zen 5, when viewed with AMD's increasingly strong Datacenter business, is sure focused on Server markets where a 30-40% energy efficiency improvement can wreak havoc on competitors.
Along with the clear focus on AVX512 performance gains, this architecture is probably more forward looking than anything else in recent AMD history.
With the higher margins on Datacenter sales, I think AMD was somewhat OK with not-so-great gains in raw desktop performance.
Messing up the communication though is on them!

 

[Darth Vader]

AMD Ryzen 5 9600X Review, Extremely BAD Value!​

 

[desiibond]

AMD Ryzen 5 9600X Review, Extremely BAD Value!​

When JayZ did his tests and reviews, he clearly mentioned that the BIOS patch he is using for review is a pre-launch patch and it may not be stable. These idiots (Hardware Unboxed), instead of trying to understand why they among all others are getting lower scores, going on and on about how bad Zen 5 is. Look at their Twitter timeline. They are trying super hard to prove that Zen5 is a disaster. Here is a screenshot from kit guru 9600X review.

Video :

 

[Darth Vader]

When JayZ did his tests and reviews, he clearly mentioned that the BIOS patch he is using for review is a pre-launch patch and it may not be stable. These idiots (Hardware Unboxed), instead of trying to understand why they among all others are getting lower scores, going on and on about how bad Zen 5 is. Look at their Twitter timeline. They are trying super hard to prove that Zen5 is a disaster. Here is a screenshot from kit guru 9600X review.

And who is to be blamed for that, the reviewers or AMD ? HUB tested as it is and shared their opinion based on current benchmark numbers. As Steve said, AMD could at least upped the core count across the board, i was really hoping we would get in the range of 8-24 Core CPU (moar core,less AI ) considering the power efficiency of Zen 5. Not gonna happen unless Intel comeback stronger with Arrow Lake.

 

[ze_cook]

When JayZ did his tests and reviews, he clearly mentioned that the BIOS patch he is using for review is a pre-launch patch and it may not be stable. These idiots (Hardware Unboxed), instead of trying to understand why they among all others are getting lower scores, going on and on about how bad Zen 5 is. Look at their Twitter timeline. They are trying super hard to prove that Zen5 is a disaster. Here is a screenshot from kit guru 9600X review.

HUB and GN are two of the top reviewers who have a credible record of basing conclusion on data, not promises, and I'd trust them far more than Jayz2c(who drilled a motherboard near the socket to mount a cooler or something) or LTT (with their trash culture, surface level knowledge and often incorrect data). For non AVX512 workloads, zen5 is a yawn, and according to HUB, compared to 7700 (non x), 9700x is 7% faster at iso power (running cinebench r24 multi). That's not nothing, but not something I'd be excited about. 9700x trades blows with the 7700x in gaming workloads, and will probably be a few % faster compared to 7700, at iso power.

Debugging performance regression isn't on HUB, that's between the mobo vendors and AMD. These people are hardware reviewers, and they're simply reviewing it at the current state of release, as you can buy. They sometimes do a look back video, comparing hardware of yesteryear, and you notice the improvement from release date.

Unstable memory, bios, agesa issues has plagued AMD during earlier releases as well, and a reviewer should not sidestep the issue just because it will (may) be fixed later on.

To give credit where it is due, these are really good at avx512 workloads, which their intel consumer counterparts don't even support, and will find their place at render farms, server clusters. Code compilation is something they do really well, and emulation too probably.

 

[variablevector]

When JayZ did his tests and reviews, he clearly mentioned that the BIOS patch he is using for review is a pre-launch patch and it may not be stable. These idiots (Hardware Unboxed), instead of trying to understand why they among all others are getting lower scores, going on and on about how bad Zen 5 is. Look at their Twitter timeline. They are trying super hard to prove that Zen5 is a disaster. Here is a screenshot from kit guru 9600X review.

View attachment 204059

Video :

There are no non-X chips on that graph.

If AMD released another 9000X chip today that had an extra 10% more performance at the cost of 50% higher power draw and then they waited another two years to put out a 10000X chip that can match it without the need for that 50% increase in power draw, would you consider that an earth-shattering, monumental achievement of AMD's engineering team?

 

[desiibond]

And who is to be blamed for that, the reviewers or AMD ? HUB tested as it is and shared their opinion based on current benchmark numbers. As Steve said, AMD could at least upped the core count across the board, i was really hoping we would get in the range of 8-24 Core CPU (moar core,less AI ) considering the power efficiency of Zen 5. Not gonna happen unless Intel comeback stronger with Arrow Lake.

AMD's major mistake is sending review units when BIOS was not ready and new chipset based boards too were not ready for launch. Sticking to 6 and 8 core chipsets even now is a mystery. God only knows why they are trying to stick to that. This is AMD's 10++++++++++ moment TBH. Even I would have had an itch to upgrade if they upped the core count. I will stick to 5800x for another year and then upgrade to 5950x. This being said, the way Hardware Unboxed is acting on their channel, on X, it look as if they wanted to go the -ve way. Steve from GN was on the point. He called out that his CPU samples had lot of issues and he has to see why they are failing.

 

[altair21]

Now, combine these two (desktop + laptop) and you will understand why its not less than 100/- per month.

Just wanted to ask one thing, are you really running your computer at peak loads 24/7 such that you'll save more than 10 units on your electricity bill? If so, your bill would be large enough that 10 units(give or take) won't matter much at all and you'll be better served setting up a small solar plant (yes I know it's a lot of investment and time upfront but it pays off massively).

For laptops. any gains on efficiency is absolutely worth top dollar since these are meant to be portable machines but for desktops, I don't think so specially at the prices amd is charging right now, maybe when these prices fall 6 months down the line it'll be worth it,definitely not right, I'd rather go all in with performance instead of minimal savings


EDIT: I agree with comments above, I'd rather have more cores for better multi processing than efficiency, AMD's current core count is a travesty considering what Intel's been giving since 12th gen

 

[desiibond]

HUB and GN are two of the top reviewers who have a credible record of basing conclusion on data, not promises, and I'd trust them far more than Jayz2c(who drilled a motherboard near the socket to mount a cooler or something) or LTT (with their trash culture, surface level knowledge and often incorrect data). For non AVX512 workloads, zen5 is a yawn, and according to HUB, compared to 7700 (non x), 9700x is 7% faster at iso power (running cinebench r24 multi). That's not nothing, but not something I'd be excited about. 9700x trades blows with the 7700x in gaming workloads, and will probably be a few % faster compared to 7700, at iso power.

My JayZ is just a reference on calling out something that is more related to FW/SW than Hardware. My go-to channels have been GN, der8auer and then I go to anandtech/tomsharware for detailed review. The less we talk about LTT the better.

Yes, Zen 5 is more and more looking like a made-for-data-center architecture. I am actually waiting for 9900X and 9950X. Also expecting a 105W TDP 9800X / 3D that should show the Zen 5 in its true performance. And of course, need to wait for stable BIOS to really know how efficient it is.

Debugging performance regression isn't on HUB, that's between the mobo vendors and AMD. These people are hardware reviewers, and they're simply reviewing it at the current state of release, as you can buy. They sometimes do a look back video, comparing hardware of yesteryear, and you notice the improvement from release date.

Unstable memory, bios, agesa issues has plagued AMD during earlier releases as well, and a reviewer should not sidestep the issue just because it will (may) be fixed later on.

This is what my point is. It is clear that there are BIOS or memory related inconsistencies. If their job is to just run benchmarks and compare with previous gen, how are they different from our chhapri mobile reviewers who read specs from box, run some benchmarks, two three games, take some photos and call it a day?

To give credit where it is due, these are really good at avx512 workloads, which their intel consumer counterparts don't even support, and will find their place at render farms, server clusters. Code compilation is something they do really well, and emulation too probably.

Yeah, refer to the Phoronix benchmarks. They were super happy the way Zen 5 works with Linux.

Just wanted to ask one thing, are you really running your computer at peak loads 24/7 such that you'll save more than 10 units on your electricity bill? If so, your bill would be large enough that 10 units(give or take) won't matter much at all and you'll be better served setting up a small solar plant (yes I know it's a lot of investment and time upfront but it pays off massively).

When I turn my desktop on (on most of the evenings and weekends), it's either gaming or working on VMs/pods. This includes two hours of daily game time for kids (each one gets 1 hour). So, yes, it is at heavy utilization whenever it is on. Will see how desktop CPU/GPU is heading towards. My RTX3090 is working perfectly. If I face any issue with it, I may ditch desktop and go for a console as I have a MacBook).

For laptops. any gains on efficiency is absolutely worth top dollar since these are meant to be portable machines but for desktops, I don't think so specially at the prices amd is charging right now, maybe when these prices fall 6 months down the line it'll be worth it,definitely not right, I'd rather go all in with performance instead of minimal savings

This is the way, right? Best time to buy new CPU is half a year or more into the current architecture.

EDIT: I agree with comments above, I'd rather have more cores for better multi processing than efficiency, AMD's current core count is a travesty considering what Intel's been giving since 12th gen

Do you know that Intel's more and more cores and pushing for more voltage to power those cores is what is causing 13th gen and 14th gen failures? Both are on two sides of the spectrum. I don't want 23984623984732 cores in my CPU. I do not like paying so much for an 8 core CPU as well. What we need is something in between. Still wondering why AMD did not ditch 6 core chiplet and go with 8 core and 12 core chipsets. Probably they wanted to keep the power consumption in check? A 24 core Ryzen would be epic though.

 

[vishalrao]

I like this KitGuruTech video... some sanity.

.

And this guy hits the nail on the head IMO - AMD is focussed on Enterprise/Datacenter market and doesn't care about individuals (client) market for CPUs and GPUs. Someone had mentioned this same thing a long while back - can't remember if it was this same guy:

.

I guess the easy solution for regular users like us is to just enable PBO and maybe EXPO settings to get the real perf uplifts compared to Zen 4.

 

[desiibond]

I like this KitGuruTech video... some sanity.

.

And this guy hits the nail on the head IMO - AMD is focussed on Enterprise/Datacenter market and doesn't care about individuals (client) market for CPUs and GPUs. Someone had mentioned this same thing a long while back - can't remember if it was this same guy:

.

I guess the easy solution for regular users like us is to just enable PBO and maybe EXPO settings to get the real perf uplifts compared to Zen 4.

Exactly what I was trying to convey. This time around, motherboards and memory will be huge differentiating factor given how the CPUs are behaving when they are paired with right motherboard and memory. One who gets it perfect is going to sell lot of units for sure.

MLID summed all this confusion pretty well in his short analysis video. His point that AMD should have waited for some more weeks, stabilize the platform and release is so true.

Gordon from PCWorld decided to delay their review as the data is very inconsistent. Isn't this what good journalism is? There are channels who wanted to be the first to publish review and they took the '-ve' route for more hits. There are channels who published review the moment the embargo lifted and they called out in the beginning itself that the results are inconsistent.

Gordon's words: As a long time hardware reviewer, there are lot of decisions that go into being right rather than being first and on-time.
In this, they also explained how they get the reference data from AMD and how AMD asks them to reach out to them if the review benchmarks are comparable.

These guys showed data for Cinebench single thread and multi thread scores. 7700X had 115 score and 9700X had 132 score in single thread. But then they both got same score in multi-thread. This indeed looks like scheduling related problem in Windows, given that Linux benchmarks show solid performance (phoronix).

 

[kiran6680]

Efficiency. The 9700x matches or beats 7700X while consuming up to 40% less power.

I love efficiency and apparently you too. But most of the market cares for efficiency only in laptops. So AMD's focus on efficiency in desktop CPUs is good for us, but maybe bad for themselves.

They lose out on laptop marketshare because of lower fabrication capacities, and no illegal monopolies like Intel has enjoyed over the decades. For laptop success, you need lots of chips, small, lower core count, and Dell/ HP in your pocket.

 

[ishashank]

These guys showed data for Cinebench single thread and multi thread scores. 7700X had 115 score and 9700X had 132 score in single thread. But then they both got same score in multi-thread. This indeed looks like scheduling related problem in Windows, given that Linux benchmarks show solid performance (phoronix).

One is 65watt part and other is 105watt part and they score equal. AMD did good here, doesn't look like a scheduling problem thus the negative reviews.

AMD has become the new intel however, look at that Indian pricing, 37k for 8 core, it is a shame. At a much lesser price you can get 7800x3d and call it a day.

 

[kiran6680]

I think that AMD marketing / PR missed the boat.
They should have specifically asked reviewers (AMD, and for that matter all HW vendors do provide a "Review Guide" along with review samples) to emphasize the energy efficiency.

How can they? The reviewers don't work for AMD, they work for the audience. The audience are apparently obsessed with gaming. Note how a good technical reviewer is even called "Gamers" Nexus. For a gamer, there is a very very low importance of CPU efficiency when their GPU is consuming over twice the power.

If a gamer is gaming all the time, excepting professional Twitch streamers, their parents are paying the power bill so they don't care. If a gamer is gaming only in spare time, a few hours a week of low power consumption by a CPU doesn't matter anyway.

When the fall of current civilization will be studied by the next civilization, they will say these guys had great hardware for the era but wasted it all on playing games - even inventing a new verb "game" instead of the already existing verb "play" for the noun "game".

 

[desiibond]

One is 65watt part and other is 105watt part and they score equal. AMD did good here, doesn't look like a scheduling problem thus the negative reviews.

AMD has become the new intel however, look at that Indian pricing, 37k for 8 core, it is a shame. At a much lesser price you can get 7800x3d and call it a day.

Not really 'much lesser price'. 7800x3d costs 3k less. But when it comes to gaming, there is no comparison. I would pick 7800x3D over the 9700X any given day, even if it costs more than the 9700x. The 7700x though costs 30.5k while 9700x costs 37k. I would go with the 9700x because the 7700x has been the bad wolf in 7xxx series. Should I buy the 9700x now? No. I would definitely wait a bit. For those already using 7xxx CPU, there is no point in upgrading. For those on AM4 5xx or older, entire platform upgrade is necessary. So, better to wait for Arrow Lake as well to release and then there will be definite price cuts across the board for AMD CPUs. We will be spoilt for choice then.

How can they? The reviewers don't work for AMD, they work for the audience. The audience are apparently obsessed with gaming. Note how a good technical reviewer is even called "Gamers" Nexus. For a gamer, there is a very very low importance of CPU efficiency when their GPU is consuming over twice the power.

If a gamer is gaming all the time, excepting professional Twitch streamers, their parents are paying the power bill so they don't care. If a gamer is gaming only in spare time, a few hours a week of low power consumption by a CPU doesn't matter anyway.

There are few big problems with the way AMD handled the release of these chips.
1. They did not wait for platform instability issues to clear.
2. These are clearly not gaming CPUs (as recently we are seeing the trend of specific type of CPUs marked as for gaming like the x3D parts) and AMD should have given clear instructions that they cannot be compared with 7800x3D. They made a boo boo by saying that the 9700x will match 9800x3D in gaming performance while being a 65W part (and 15% IPC improvement). So reviewers latched on to that and expected huge performance improvement.
3. AMD trumpeted efficiency when they talked about Strix Point for laptops but they trumpeted about performance of desktop counterparts. They should have played efficiency angle for these two chips and trumpet about performance of their flagship chips and the upcoming x3D parts.

 

[kiran6680]

There are few big problems with the way AMD handled the release of these chips.
1. They did not wait for platform instability issues to clear.

Right, that was embarrassing for them. Though for me as a consumer, AMD better have problems now than have problems after selling chips and the chips breaking down.

2. These are clearly not gaming CPUs (as recently we are seeing the trend of specific type of CPUs marked as for gaming like the x3D parts) and AMD should have given clear instructions that they cannot be compared with 7800x3D. They made a boo boo by saying that the 9700x will match 9800x3D in gaming performance while being a 65W part (and 15% IPC improvement). So reviewers latched on to that and expected huge performance improvement.

A. I'm not sure AMD can dictate what can be compared against what. Freedom of expression etc.
B. Even file compression/decompression/puget/blender show 9700X very similar to 7700X, which is embarrassing. So it's not just gaming that's the problem. I know if optimised in BIOS, 9xxx might be much better, but out-of-box comparison cannot be prevented.

3. AMD trumpeted efficiency when they talked about Strix Point for laptops but they trumpeted about performance of desktop counterparts. They should have played efficiency angle for these two chips and trumpet about performance of their flagship chips and the upcoming x3D parts.

Ok, I didn't know that. But AMD sells a small fraction of laptop chips as compared to Intel, for various reasons like fab capacity, prioritising server parts, Dell/HP owning corporate market and Intel owning them etc. So AMD talking about laptops will make sense only after they can fix those fundamental problems.

 

[tantrikt]

One of the biggest gripes with amd chips was single core performance. Most reviews of intel vs amd have that sentence always. So far without new optimised and tailored chipsets for zen 5 they are performing satisfactorily on that front, with unchanged multi core performance and that too at lower TDP. With zen5 it may not be a giant leap in performance across the board but as some reviews point it it's more efficient and wholesome.
At any rate zen 5 needs more time to sink in and more testing. The motherboard and ram ecosystem around it maybe needs an update to unlock true potential of these chips.
The pricing will also stabilise near the time arrow lake comes out.