First professional review of Intel's Rocket Lake (11700K) is out.. and it's a disaster

[LAwLz]

1 hour ago, RejZoR said:

I never thought we'd come to this point where Intel is just outright terrible. I thought Intel would catch up by this time, not get even worse.

Don't listen to the people on LTT. Read the review for yourself instead. Rocket Lake isn't worse than Skylake. People are cheery picking some benchmarks and pretending like that's the overall score.

Overall Rocket Lake is better than Skylake. Zen 3 is still better than anything Intel has to offer though.

 

This is the conclusion Anandtech came to:

Quote

All workloads at their core, even when browsing the web or word processing, can be split into integer (whole numbers, most workloads) and floating point (numbers with decimal places, workloads with math). In our testing, we saw the following:

• Single thread floating point: +19.0%
• Multi-thread floating point: +19.5%

Sounds great, right?

• Single thread integer: +13.0%
• Multi-thread integer: +7.3%

Oh. While Intel’s claim of +19% is technically correct, it only seems to apply to math-heavy workloads. The benefits of non math-based throughput are still better than average, 7-13%, but vary rarely do Intel’s big claims come with an easily identifiable asterisk.

When we look at our real-world data, in almost every benchmark the 11700K either matches or beats the 10700K, and showcases the IPC gain in tests like Dolphin, Blender, POV-Ray, Agisoft, Handbrake, web tests, and obviously SPECfp. It scores a big win in our 3DPM AVX test, because it has AVX-512 and none of the other CPUs do.

 

 

It is quite frankly astonishing that people on this forum, who claim to be tech enthusiasts, are so moronic that they aren't read benchmarks or apparently can't even be bothered to read the conclusion of the article. They seem to find it more enjoyable to be fanboys and throw shit at brand X than actually digging into the tests.

It performs worse in like 2 out of ~40 tests? Lol, let's say it performs worse and that Intel are going backwards! That's fun, right?

Or maybe some people on this forum missed that quite a few of the charts are "lower is better" and think that "higher number is better" because they can't even be bothered to read what the test is or what it does. I dunno. It is quite sad and infuriating to see such levels of stupidity from people though.

 

Overall, Rocket Lake is about 7-13% better than Skylake, with some scenarios offering better uplift than that.

[Nowak]

29 minutes ago, LAwLz said:

I assume you mean the AVX-512 results. This is what performance looks like with that power draw.

 

Twice the power usage, but almost 6 times the performance.

 

The chip is not great. The 5800X is a much better chip overall. My point is that we don't have to lie or exaggerate about how bad the 11700K is.

I can't tell if some people in this thread want Intel to fail and therefore exaggerate stuff, or if they are just dumb and don't understand what the graphs they are looking at shows.

I definitely fall into the latter camp, as I currently use Intel in both my laptop and desktop. I've been thinking about switching to AMD for a while, but I'm still undecided.

 

Ah, I'm still at least a year or two out from needing an upgrade. I'll see what Intel and AMD offer in 2022 or 2023.

[LAwLz]

18 minutes ago, Nowak said:

I definitely fall into the latter camp, as I currently use Intel in both my laptop and desktop. I've been thinking about switching to AMD for a while, but I'm still undecided.

 

Ah, I'm still at least a year or two out from needing an upgrade. I'll see what Intel and AMD offer in 2022 or 2023.

Like I said a couple of pages ago, I think it is most certainly the best decision to not buy anything right now.

Not only because of really messed up prices and availability, but also because of things like DDR5 being just around the corner. Buying the second generation of DDR5 compatible CPUs will most likely be a better long term deal than buying the last gen of DDR4.

The price premium of DDR5 will have gone down, and the higher rated, better DDR5 modules will have hit the market.

 

It has historically been a bad decision to buy the first or last generation of processors for a new DDR standard. Right now we are at the last generation of DDR4, and this year or next year we'll get the first generation of DDR5.

[Nowak]

2 minutes ago, LAwLz said:

Like I said a couple of pages ago, I think it is most certainly the best decision to not buy anything right now.

Not only because of really messed up prices and availability, but also because of things like DDR5 being just around the corner. Buying the second generation of DDR5 compatible CPUs will most likely be a better long term deal than buying the last gen of DDR4.

The price premium of DDR5 will have gone down, and the higher rated, better DDR5 modules will have hit the market.

 

It has historically been a bad decision to buy the first or last generation of processors for a new DDR standard. Right now we are at the last generation of DDR4, and this year or next year we'll get the first generation of DDR5.

So I shall hold onto Coffee Lake until 2023, got it.

[LAwLz]

6 minutes ago, Nowak said:

So I shall hold onto Coffee Lake until 2023, got it.

My recommendation would be that unless you for some reason absolutely need a new processor right now (your current one is broke or you need a faster CPU for work related things) then yes, wait.

First gen DDR5 should be out in late 2021 or early 2022 (with Alder Lake and Zen 4), so by late 2022 to mid 2023 we should have decent or good DDR5 prices, and better than minimum-spec DIMMs on the market.

[illegalwater]

1 hour ago, LAwLz said:

If I look through your post histories, will I find some posts where you two bash AMD in that case? Because Ryzen 5000 also perform worse than Ryzen 3000 in some specific metrics.

For example the L3 latency has gone up from Zen to Zen 2 and to Zen 3 again. Latency also went up for a lot of x87 instructions, and some instructions have lower per clock throughput in zen 3 compared to zen 2, such as RORX and BZHI.

 

 

CPUs are very complex, and it is often worth losing performance in some areas if you can better allocate resources to improving performance in other situations. For example higher latency but higher overall throughput was a design choice AMD made with zen 3 for some instructions (such as a lot of x87). If you only look at "it is worse in this areas" and use that to paint some generalized image then it is very easy to get the wrong impression. But I feel like some people want to get the wrong impression...

The key differences here are that Zen 3 didn't take over 5 years to happen, and Zen 3 was faster across the board in consumer applications. Pretty sure Zen 3 doesn't run slower than Zen 2 in any games for example.

 

Also I find it telling you're having to dig up some minor performance regressions for Zen 3 that only really manifest on paper, meanwhile I'm basing my opinion of Rocket Lake on how it performs in actual tests.

[illegalwater]

1 hour ago, LAwLz said:

It is quite frankly astonishing that people on this forum, who claim to be tech enthusiasts, are so moronic that they aren't read benchmarks or apparently can't even be bothered to read the conclusion of the article. They seem to find it more enjoyable to be fanboys and throw shit at brand X than actually digging into the tests.

It performs worse in like 2 out of ~40 tests? Lol, let's say it performs worse and that Intel are going backwards! That's fun, right?

Or maybe some people on this forum missed that quite a few of the charts are "lower is better" and think that "higher number is better" because they can't even be bothered to read what the test is or what it does. I dunno. It is quite sad and infuriating to see such levels of stupidity from people though.

 

Overall, Rocket Lake is about 7-13% better than Skylake, with some scenarios offering better uplift than that.

Two tests? Dude look at the benchmarks again, it's worse or at best tied with Skylake in most of the game benchmarks, and in several of the other benchmarks it ties with the 9900K. If that's all Intel can muster after five years I'd consider that pretty pathetic.

 

I don't know why you're so hellbent on fanboying over a multibillion dollar corporation.

[RoseLuck462]

AMD wins again!

[Nowak]

1 hour ago, LAwLz said:

My recommendation would be that unless you for some reason absolutely need a new processor right now (your current one is broke or you need a faster CPU for work related things) then yes, wait.

First gen DDR5 should be out in late 2021 or early 2022 (with Alder Lake and Zen 4), so by late 2022 to mid 2023 we should have decent or good DDR5 prices, and better than minimum-spec DIMMs on the market.

I thank you for the help, but we also digress.

[D13H4RD]

3 hours ago, illegalwater said:

Two tests? Dude look at the benchmarks again, it's worse or at best tied with Skylake in most of the game benchmarks, and in several of the other benchmarks it ties with the 9900K. If that's all Intel can muster after five years I'd consider that pretty pathetic.

 

I don't know why you're so hellbent on fanboying over a multibillion dollar corporation.

The thing though is that due to all the 10nm woes, Intel’s engineers have also spent a lot of those 5 years polishing the absolute heck out of the 14nm Skylake-based architecture to the point where it would be a herculean task for the 10nm-based parts to even come close to the clockspeed and overall desktop performance it could muster. Though the fact they backported the Sunny Cove cores from Ice Lake on 14nm does kind of show that Skylake was already running into hard limits. 

 

Not defending them here, but the general reaction to Rocket Lake well before Intel showcased seems to be a collective “meh”, because we knew Alder Lake is coming soon with a lot of changes and Rocket Lake is really just a gap-closer in a sense. Unless one needs AVX-512 or PCIe Gen4 in an Intel platform right now, Comet Lake or even Coffee Lake’s still more than fine.

 

Everyone’s already said it; Rocket Lake is a bad release relative to what we’ve seen in the last few. There’s no need to keep beating a dead horse when the horse itself is already dying before it even reached the racetrack. Is it a bad product? Not necessarily but it’s not compelling enough for people to just not wait for Alder Lake/Zen 4 or buy discounted Comet Lake parts when they really need something from Intel right now.

[MightyMishka]

On 3/5/2021 at 1:38 PM, spartaman64 said:

i saw some rumors that the 11900k hits 98C with a 360mm AIO 

Jesus christ, wow.

[MightyMishka]

I feel bad for Intel, AMD is crushing them right now, even with the silicon shortages and stock problems.

[Brooksie359]

On 3/5/2021 at 4:33 PM, Kilrah said:

Well, rockets do get quite hot...

And consume a massive amount of energy to work. 

[RejZoR]

8 hours ago, LAwLz said:

Don't listen to the people on LTT. Read the review for yourself instead. Rocket Lake isn't worse than Skylake. People are cheery picking some benchmarks and pretending like that's the overall score.

Overall Rocket Lake is better than Skylake. Zen 3 is still better than anything Intel has to offer though.

 

This is the conclusion Anandtech came to:

 

 

It is quite frankly astonishing that people on this forum, who claim to be tech enthusiasts, are so moronic that they aren't read benchmarks or apparently can't even be bothered to read the conclusion of the article. They seem to find it more enjoyable to be fanboys and throw shit at brand X than actually digging into the tests.

It performs worse in like 2 out of ~40 tests? Lol, let's say it performs worse and that Intel are going backwards! That's fun, right?

Or maybe some people on this forum missed that quite a few of the charts are "lower is better" and think that "higher number is better" because they can't even be bothered to read what the test is or what it does. I dunno. It is quite sad and infuriating to see such levels of stupidity from people though.

 

Overall, Rocket Lake is about 7-13% better than Skylake, with some scenarios offering better uplift than that.

By "getting worse" I didn't mean by regressing in performance, but not catching up to AMD at all in all this time. Given how much time has passed since first Ryzen and Intel is still just struggling to release even underwhelming products is what is baffling me. It's not like they learned they kinda suck last year. The moment AMD dropped first generation Ryzen they should know something is up. And what happened since? Big fat juicy nothing. Even more so because they want you to buy another stupid motherboard once again. I could see people being enthusiastic about Rocket Lake if you could drop the Rocket Lake chip into same motherboard you already have. Being required to swap it all you may just as well just buy Ryzen 5800X and have a better CPU overall on a better platform overall. Unless you really need something very very very specific from Intel. And that's by 5800X being available for few months now. 11700K was just released. Imagine being late to the party and not offering anything exciting.

 

Hell, there was even a time when I was genuinely excited when I heard about Sunny Cove years ago and how it should give Intel back the competing edge. And this is it? Ugh?! In fact it was Skylake that I was super excited back then when I was reading leaks it should come with exciting new architecture with extra built in eDRAM cache that should significantly uplift the performance. And even that ended up just being the damn same CPU that did end up having eDRAM built in, in one mid end model and only being used for the iGPU in a similar way AMD's Infinity Cache works. Outside of iGPU, that eDRAM was a dead weight. I was so disappointed I bought their last gen HEDT with more cores instead at even slightly lower price (5820K instead of 6700K). Intel has been rather underwhelming from back then already...

[SupaKomputa]

I was hoping Intel would be competitive to push Ryzen 5000 price down.

But looks like it wont happen in the near future.

The only card they can pull is having them at a lower cost.

Even that still won't put them in a good position, as you have to buy more PSU watt and better than average cooling.

Sad, really sad.

[Kotonoha]

9 hours ago, LAwLz said:

Like I said a couple of pages ago, I think it is most certainly the best decision to not buy anything right now.

Not only because of really messed up prices and availability, but also because of things like DDR5 being just around the corner. Buying the second generation of DDR5 compatible CPUs will most likely be a better long term deal than buying the last gen of DDR4.

The price premium of DDR5 will have gone down, and the higher rated, better DDR5 modules will have hit the market.

 

It has historically been a bad decision to buy the first or last generation of processors for a new DDR standard. Right now we are at the last generation of DDR4, and this year or next year we'll get the first generation of DDR5.

i built my x79 platform during ddr4 transition, it will outlive ddr4 platform and i will move to ddr5 platform launches

[D13H4RD]

2 hours ago, RejZoR said:

I could see people being enthusiastic about Rocket Lake if you could drop the Rocket Lake chip into same motherboard you already have. Being required to swap it all you may just as well just buy Ryzen 5800X and have a better CPU overall on a better platform overall. 

You can use RKL-S on Z490 boards. MSI even confirmed PCIe Gen4 compatibility, confirming long-standing theories that some Z490 boards already have PCIe Gen4 supported, just not active. 

 

The real issue is that unless you really need PCIe Gen4 and/or AVX-512 instructions, those already on Z490 don't really have a tangible reason to step up, especially since Comet Lake-S as it stands is still a perfectly capable generation. 

[Delicieuxz]

11 hours ago, LAwLz said:

I assume you mean the AVX-512 results. This is what performance looks like with that power draw.

 

Twice the power usage, but almost 6 times the performance.

 

The chip is not great. The 5800X is a much better chip overall. My point is that we don't have to lie or exaggerate about how bad the 11700K is.

I can't tell if some people in this thread want Intel to fail and therefore exaggerate stuff, or if they are just dumb and don't understand what the graphs they are looking at shows.

 

Going by just the AVX2 results from this chart, I don't think people are exaggerating when they say the 11700k's power-draw is awful compared to AMD.

 

[D13H4RD]

36 minutes ago, Delicieuxz said:

 

Going by just the AVX2 results from this chart, I don't think people are exaggerating when they say the 11700k's power-draw is awful compared to AMD.

 

Well, it's not so much about that. Everyone knows Intel hasn't had a fantastic time when it comes to managing peak power draw. 

 

It's just that everyone is laser-focused on the figure that shows it drawing almost 300W on AVX-512 when the figure of it drawing 224W on AVX2 is more indicative of what most workloads would see. 

 

To be clear, it is still not great, especially when compared to AMD's miserly power draw. It's just not as bad as what the echo chamber would suggest. 

[Rym]

Well it's quite disappointing to see the new cpu performance to be honest.

 

Yes it works well in avx 512 loads but most ppl judge it by it's gaming performance which is sad to say the least.

 

I understand  it's a stopgap product however the fact is it costs almost 500 euro for the i7 and over 600 euro for the i9 currently while providing the same or less performance than a 193 euro CPU (10600KF overclocked) in gaming.

 

I also understand people hope for Alder Lake to be the new big thing but it doesn't look like people fully understand the fact that Alder Lake is a first generation CPU on a very different architecture for PCs with half cores being good performance and the other half bad performance with good power efficiency.

 

I would not be surprised if Alder Lake flops as well but personally competition between Intel and AMD is good because in a war between 2 companies to release the best product there is only 1 winner, the consumer. Less price, more performance.

 

Moving forward, the current rumours are that Zen 4 is another gigantic leap forward providing up to 40% more performance and a 25% IPC uplift compared to Zen 3. I'd like to take these rumours into consideration as well when we consider Alder Lake, as really Alder Lake should realistically be compared to Zen 4 in my opinion.

 

Furthermore AMD is fighting a war on two fronts and seems to be pushing the competition hard into their home turf, with their RDNA 3GPUs moving onto an all-new design which in internal testing has seen the performance target to be 2.5x that of the 6900 XT. While AMD may overwhelm Nvidia's GPUs with raw power, Nvidia Lovelace might also provide strong competition due to it being 3rd generation refined RTX and having well developed DLSS. AMD currently did not release their DLSS competitor and their ray tracing performance is not up to par with RTX 30 series while the raw power of their GPUs does compete and sometimes overwhelm RTX 30 series GPUs, by up to a 50+ FPS advantage in select games (Battlefied if I remember correctly).

 

 

It's looking to be a wild ride from here on out.

[LAwLz]

2 hours ago, Delicieuxz said:

 

Going by just the AVX2 results from this chart, I don't think people are exaggerating when they say the 11700k's power-draw is awful compared to AMD.

 

Intel draws more power. Nobody is debating against that. 

The problem i have are the people now saying Intel uses 300 watts because they are looking at the AVX512 numbers. 

 

People are reaching the right conclusion (Intel uses more power) but to the wrong degree (they say 300 watts instead of 230 watts). 

 

It's important to get facts correctly. Otherwise it will be a clusterfuck when AMD adopts AVX512, when Intel releases the next gen CPUs and the numbers seem to be too big of an improvement, our when others start posting reviews and their numbers are very different. 

[illegalwater]

10 hours ago, D13H4RD said:

The thing though is that due to all the 10nm woes, Intel’s engineers have also spent a lot of those 5 years polishing the absolute heck out of the 14nm Skylake-based architecture to the point where it would be a herculean task for the 10nm-based parts to even come close to the clockspeed and overall desktop performance it could muster. Though the fact they backported the Sunny Cove cores from Ice Lake on 14nm does kind of show that Skylake was already running into hard limits. 

 

Not defending them here, but the general reaction to Rocket Lake well before Intel showcased seems to be a collective “meh”, because we knew Alder Lake is coming soon with a lot of changes and Rocket Lake is really just a gap-closer in a sense. Unless one needs AVX-512 or PCIe Gen4 in an Intel platform right now, Comet Lake or even Coffee Lake’s still more than fine.

 

Everyone’s already said it; Rocket Lake is a bad release relative to what we’ve seen in the last few. There’s no need to keep beating a dead horse when the horse itself is already dying before it even reached the racetrack. Is it a bad product? Not necessarily but it’s not compelling enough for people to just not wait for Alder Lake/Zen 4 or buy discounted Comet Lake parts when they really need something from Intel right now.

I guess I'm just baffled they even bothered releasing this thing. They would've been better off just taking the L for a year while Alder Lake finishes, especially since the gap between the two is seemingly going to be less than six full months. All Rocket Lake does is further the perception that Intel has become a joke in the desktop market, it's also negatively impacted the last thing Intel had going for them (gaming performance).

 

I mean could you imagine if Nvidia went five years just releasing refreshes of Pascal on 14nm, and then they finally come out with a new architecture.. and the best thing you could say about it is that its just marginally better in some areas? I'm not even going to mention performance regressions in this analogy because apparently that ticks people off because "CpUs ArE CoMpLeX" even though there have been countless releases throughout history that were across the board better in real world tasks than their predecessors.

 

I feel for the engineers that had to waste their time on this.

15 minutes ago, comander said:

Speculation - internal cache timings suck. 

 

Intel made the uarch on the assumption that it would be used with 10nm. 

 

14nm, naively, is 40% longer. If you have a block of cache, getting an electron from one end to the other will take 40% longer if you're going left/right. If you're doing manhattan distance from opposite ends of the block it's +80% relative to 10nm. 

 

If you hadn't to slow down timings so that the electrons have time to move across the cache... Well... There goes most of your IPC uplift. 

 

It's possible that clocking the design to the moon will help a bit but...

Just another reason why Rocket Lake should've never been released. Low latency was one of Intel's big strengths, and they blew it here.

[AndreiArgeanu]

I feel like 11th gen was made just for the sake of having at least 2 generation of cpu's on this socket before they move to whatever else they've got. Still 225W power consumption in POV-Ray and as high as 290W in AVX512 is quite ridiculous. And from the looks at gaming performance, AMD may not have to lower their prices, let's hope that maybe a BIOS update will improve gaming performance at least. Regardless, things don't look so great at the moment for Intel.

[WhitetailAni]

25 minutes ago, comander said:

Beyond that the flagship is the 10900k not the 10700K. 

The 10900K is also 8C/16T, and from my understanding, will just be a cherry-picked 10700K.

[D13H4RD]

8 minutes ago, AndreiArgeanu said:

And from the looks at gaming performance, AMD may not have to lower their prices...

That's the disappointing part.

 

I really, really hope Intel doesn't have the nerve to insist on charging a premium over the already significant premium AMD has put on Zen 3 for RKL-S, for a product that, so far, looks to be the inferior option in performance outside of some fringe cases like AVX-512.

 

Before Zen 3, Intel had a point to why its products tend to cost a bit more than the equivalent Ryzen; they were the best choice for gaming. Now though, I don't think even Intel's reliability factor is significant enough to warrant paying more for their products at this point.