Intel Core i7-11700K beats Ryzen 9 5950X by 8% in Geekbench 5 single-core benchmark

[MartinPrince]

So, who is right and who is not?

It seems your Core i7-9700K with 8 threads and no HT is not faster than the 32-thread Ryzen 9 3950X.

An app which can utilise only 10 threads will unleash more performance than your old CPU.

If English is not your first language then I apologise but your level of comprehension seems to be sorely lacking. I'll try and break it down into small segments so you understand the topic. Here we go again...

• We are talking about single threaded performance or maybe more accurately apps that can't take full advantage of multicore CPU's
• We covered this already last year in this thread https://www.overclockers.co.uk/forums/threads/amd-vs-intel-single-threading.18872467/page-3
• In one of the apps (DXO Photolab 3) I used my old and 'only' 8 core/ 8 thread 9700K which was proved to be faster than a 24 core/48 Thread 3960X.

Here is a rough table of results from that thread. You seem to have forgotten that you and people like humbug where under the wrong impression that a 3600 with it's 6 core/12 threads would be faster than my 'only' 8 core/8 thread CPU in this 'not fully multithreaded' program. As you can see humbugs 3600 was ~35% slower.

Intel [email protected] 15secs Robert896r1
Intel [email protected] 15secs MartinPrince
ThreadRipper 3960X 18secs amigafan2003
Ryzen 3900X 18secs MartinPrince
Threadripper 2920X 22secs sandys
Ryzen 3600 24secs humbug
Intel [email protected] 24secs MartinPrince
Ryzen 2700@4Ghz 25secs CAT-THE-FIFTH(mate)
Ryzen 2600 27secs CAT-THE-FIFTH
Intel 4790s 38secs MartinPrince
Intel [email protected] 44secs MartinPrince

This is in an app that unless in batch mode used about 8 cores/threads and as you can see the 9700K is faster than a 3960X never mind a 3950X. (DXO Photolab 4 can now use the GPU which can be twice as fast as the CPU depending on your GPU)

Once again just in case you are still clueless, we are talking about actual SINGLE THREADED performance as shown in real world applications not synthetic benchmarks which can be wrongly skewed to either AMD or Intel.

So to answer your question, you are clearly wrong whether you can comprehend or admit that.

 

[rustyk21]

I think that it's only a matter of time before they make use of an arbitrary number of cores. They are unlikely to scale absolutely horizontally, as they are so tightly bound to pumping out frames as fast as they can, but by the time developers have 8C/16T to play with, it would seem odd to me (as a developer) to target a specific number.

Fair enough, we definitely don't agree on that then! I only mention 8/16 because it's a common config and from a software development point of view it's incredibly hard (and expensive) to parallelize games. Even though GPU drivers etc. can be multithreaded, all else being equal, single threaded performance is king.
I don't believe developers target a specific number, so it's already abitrary, hence my comments about number of cores being irrelevant (from a holistic point of view).

I think it's fair to say that since Zen2 that was a very tenuous claim. That's all.

We'll have to agree to disgree on that too then, since it's not tenous, it's backed up by data and it's repeatable. Hence all the excitement in some circles about Zen3.
If all you want is cores and threads then Zen2 is still a good value proposition given the price and availbility differences.

 

[Jimmy Weirdarms]

We'll have to agree to disgree on that too then, since it's not tenous, it's backed up by data and it's repeatable. Hence all the excitement in some circles about Zen3.

I'd like to see some of these benchmarks that are conclusively pro-intel at resolutions that actually matter (4k, 1440p). From what I can tell it's in the realm of a couple of fps in most games rather than anything noticeable, particularly if you look at the 9900k which was the competition at the time.

 

[~>Dg<~]

both amd and intel will use pro bias benchmarks or other people. the difference is to look inbetween the lines and not be company biased yourself. being pro one team means you are actually limiting yourself.

ashes of singularity for amd is a prime eg.

 

[KungFuSpaghetti]

I think if I swapped my 3900x out for an overclocked 10900k, 11900k, 5800x, 5900x or 5950x and continued playing the games I currently play... I'd not even notice.

 

[Jimmy Weirdarms]

both amd and intel will use pro bias benchmarks or other people. the difference is to look inbetween the lines and not be company biased yourself.

Exactly! The whole "best for gaming!" thing is silly, and seems to have been a facet of team blue trying to find a marketing line to cling on to so they could maintain an image of being top of the market.

Personally I enjoy gaming on both my Zen 3 and i7 9700 systems. The Ryzen really shines on non-gaming tasks. For games I'd be hard pushed to see the difference.

 

[~>Dg<~]

I think if I swapped my 3900x out for an overclocked 10900k, 11900k, 5800x, 5900x or 5950x and continued playing the games I currently play... I'd not even notice.

some people dont some do. depends also on the game and resolutions.

 

[rustyk21]

I'd like to see some of these benchmarks that are conclusively pro-intel at resolutions that actually matter (4k, 1440p). From what I can tell it's in the realm of a couple of fps in most games rather than anything noticeable, particularly if you look at the 9900k which was the competition at the time.

So what are you asking to see exactly? The data is all out there.
I just noticed you made a comment about 'team blue' below this. That suggests to me that you are looking at this with some preconceptions baked in.
If you use a pc primarily for gaming, then why wouldn't you buy the best cpu for gaming?
Like I've consistently said right from my first post, this goes a lot further than 1080p average framerates, but you need to do your own research.

 

[KungFuSpaghetti]

some people dont some do. depends also on the game and resolutions.

I game at 1440/1080p ultrawide these days, but all my games run over my refresh rate anyway. Either way it looks like it will be a battle of improvements slightly above margin of error, in a blind taste test no one would be able to objectively tell which system they are using.

If these new Intel's end up quicker it's going to be negligible difference. It should hopefully force AMD to revisit their pricing on their 8 and 6 cores, which are slightly overpriced... but then people want them now as they are the fastest gaming chips and if you already bought into the eco system it's easier to justify shelling out for a new CPU and not a motherboard as well.

Intel's struggles will likely be worse once AM4 is superseded with the new socket. It will be an even harder sell if Intel stick to forcing a new motherboard chipset each go around whilst AMD you get a few generations.

 

[~>Dg<~]

I game at 1440/1080p ultrawide these days, but all my games run over my refresh rate anyway. Either way it looks like it will be a battle of improvements slightly above margin of error, in a blind taste test no one would be able to objectively tell which system they are using.

If these new Intel's end up quicker it's going to be negligible difference. It should hopefully force AMD to revisit their pricing on their 8 and 6 cores, which are slightly overpriced... but then people want them now as they are the fastest gaming chips and if you already bought into the eco system it's easier to justify shelling out for a new CPU and not a motherboard as well.

Intel's struggles will likely be worse once AM4 is superseded with the new socket. It will be an even harder sell if Intel stick to forcing a new motherboard chipset each go around whilst AMD you get a few generations.

i can tell. as said it depends if you can " feel " the difference. as we used to say to you for eg in pubg. amd ryzens were terrible compared to intel cpus. the fps difference was night and day difference. 30 - 50 fps sometimes in drops which in certain games instances is very noticeable. it also depends how games are benchmarked most are single player short sequences which dont even replicate how people play.

whats funny is intel said they had nothing new until 2021 for amd. that was like 2 years or more ago. yet the current line up still is often still faster easier to use in games than any amd cpus out now. you drop em in the just work. try that with amd cpus.

before the next comment they fine. as i said a few months ago check cpu forums on issues with having amd cpus. been loads of various issues and having to pay way over the odds they just dont make sense.

 

[MartinPrince]

..before the next comment they fine. as i said a few months ago check cpu forums on issues with having amd cpus. been loads of various issues and having to pay way over the odds they just dont make sense.

I would take issue with that last statement as for many people they make perfect sense. There will generally be problems with new CPU's (like not hitting stated boost clocks), though after an initial few months those tend to be ironed out and it's generally plain sailing there on out. This was the case when I got a 3900X close to launch day.

As a counter to your point I had issues with my Intel 9700K nearly a year after it came out and found I had lost performance and could not pin point where. It turned out to be the latest official bios. To get back my previous performance I had to use a modded bios using the older and faster micro-codes. (which I'm still having to use now)

I love the efficiency of the current Ryzen CPU's. My 3950X runs silent on a D15 and barely touches 70c under load. I adore silent computing. (after playing for years in band I've learnt not to take silence for granted! ).

With the Ryzen 5000 series you have the best of all worlds, finally properly authentic single thread speeds. The usual great fully threaded speeds, plus great efficiency.

Only to the most die hard of Intel enthusiasts would that not make sense...

 

[V F]

Only to the most die hard of Intel enthusiasts would that not make sense...

Who in their right mind would not want a silent computer...

 

[KungFuSpaghetti]

i can tell. as said it depends if you can " feel " the difference. as we used to say to you for eg in pubg. amd ryzens were terrible compared to intel cpus. the fps difference was night and day difference. 30 - 50 fps sometimes in drops which in certain games instances is very noticeable. it also depends how games are benchmarked most are single player short sequences which dont even replicate how people play.

whats funny is intel said they had nothing new until 2021 for amd. that was like 2 years or more ago. yet the current line up still is often still faster easier to use in games than any amd cpus out now. you drop em in the just work. try that with amd cpus.

before the next comment they fine. as i said a few months ago check cpu forums on issues with having amd cpus. been loads of various issues and having to pay way over the odds they just dont make sense.

30-50fps you must mean Ryzen 1000 series, as the 1700x was barely any better than my old Xeon except for minimum frame rates. The 2700x I had was absolutely fine though at 1080p, always ran over 100fps at worst on pube-g. The generational improvement from each round of CPU's was pretty significant when it came to gaming, but the 2000 series never competed with intel in that regard, the 8700k or whatever was out at the time was always quicker by a healthy margin. The 3000 series however closed the gap to slightly outside of margin of error for the most part.

I feel like you're comparing issues with Ryzen 1000 though, as friends that bought into those in the early days had a lot of issues with RAM, but again, XMP profiles are for intel and that's changed now. I never had a single issue on my 2700x, I didn't even need to re-install windows after swapping out all my intel parts. Plugged it in, set XMP and that was it, booted up and into windows after a short delay for hardware config. I've built 2 intel systems and about 4 Ryzen systems this year for other people and not had a single issue with any of them, but had to do the exact same things on both, which was set XMP for the ram.

Price wise yeah it's not ideal when an 8 core is costing £440, but that's IF you can get one, so they are selling regardless and that could largely be in part where people like me look at an upgrade and can justify it, because we already own a motherboard that works with it. I bought my x470 when the 2700x came out and had it ever since, so that's 3 CPU generations making it an easier pill to swallow. Fact is, on Intel if you wanted 3 consecutive generations of CPU upgrades, chances are you'd also need 3 different motherboards over that timespan. £440 on the 5800x only seems poor value if you need to buy a motherboard for it to work.
I also don't think anecdotal forum issues points to evidence of widespread issues either, people generally use the internet for help, not to post up "My PC booted up fine today" .

 

[timorous]

I game at 1440/1080p ultrawide these days, but all my games run over my refresh rate anyway. Either way it looks like it will be a battle of improvements slightly above margin of error, in a blind taste test no one would be able to objectively tell which system they are using.

I bet if the games included some late game Paradox Grand Strategy like Stellaris or whatever you could tell from the tic rate. FPS would be absolutely fine but that is not what matters in these kinds of games.

On my 2200G the reason I end a Stellaris game is usually because the tic rate slows so much progress becomes tedious and that also stops me from playing on the largest maps with the most amount of other civs because that also slows the game down. Same thing happens in Civ 6 late game where FPS is fine (in the strategy view, don't use the normal view) but turn time late game gets tedious as does troop movement as you get the occasional hitch when it needs to calculate stuff.

I really wish more of these kinds of games were included in CPU reviews because they are the few games where CPU > GPU in the buying priority can actually make sense provided the GPU can hit 30 fps + at whatever resolution you want.

 

[KungFuSpaghetti]

I bet if the games included some late game Paradox Grand Strategy like Stellaris or whatever you could tell from the tic rate. FPS would be absolutely fine but that is not what matters in these kinds of games.

On my 2200G the reason I end a Stellaris game is usually because the tic rate slows so much progress becomes tedious and that also stops me from playing on the largest maps with the most amount of other civs because that also slows the game down. Same thing happens in Civ 6 late game where FPS is fine (in the strategy view, don't use the normal view) but turn time late game gets tedious as does troop movement as you get the occasional hitch when it needs to calculate stuff.

I really wish more of these kinds of games were included in CPU reviews because they are the few games where CPU > GPU in the buying priority can actually make sense provided the GPU can hit 30 fps + at whatever resolution you want.

Never played Stellaris (but will check this out as I do love a strategy game!), but I have 1045.5 hours on Civ 6 and never had any performance related issues with it online or offline. Chugs a bit on my laptop, but not really fair to compare a mobile intel with iGPU to a 12 core desktop CPU.

 

[timorous]

Never played Stellaris (but will check this out as I do love a strategy game!), but I have 1045.5 hours on Civ 6 and never had any performance related issues with it online or offline. Chugs a bit on my laptop, but not really fair to compare a mobile intel with iGPU to a 12 core desktop CPU.

Well if FPS is the issue on the laptop just use the strategy map, looks good and FPS is far better. The issue I have with Civ 6 on my machine is that late game I get hitches when scrolling the map so I tend to avoid the largest maps and stick to around 7 other civs max.

I just wish there was more data on how fast these games run on different CPUs because the 5900X seems to do a good chunk better than the 5600X in the Civ 6 Gathering Storm AI test but with only 2 reviewers that tested it (GN and OC3D) it is not a firm conclusion. Also the AMD vs Intel performance was mixed, OC3D had Intel/AMD neck and neck where as GN had AMD well ahead. Could be memory settings on that one but when GN did their memory tuning video they did not include Civ 6 AI turn time so you can't see what the gain will be from memory tuning in that game. Then you have the Paradox Grand Strategy games and nowhere tests them, or Cities Skylines and other large scale games that have a lot of AI running in the background so again there is no data to support a buying decision either way and like I said these games are the ones where spending the extra over the 5600x / 11600k for the 8c + parts might be worth it if you get tangible tic rate improvements.

 

[Roadrunner_668]

whats the price difference though ?

 

[chrcoluk]

The PC market has gone down a dark path, instead of new products every few years which have 50%+ performance improvements, we getting more frequent release chasing to hold a tiny lead on the market, this isnt good news for the consumer. Human nature wants the latest and greatest, stock never is allowed to build up and become cheaper so is a market that will generate high demand and prices.

 

[Roadrunner_668]

its been that way for a long time now just look at grapics cards its ridiculous !

 

[KungFuSpaghetti]

Well if FPS is the issue on the laptop just use the strategy map, looks good and FPS is far better. The issue I have with Civ 6 on my machine is that late game I get hitches when scrolling the map so I tend to avoid the largest maps and stick to around 7 other civs max.

I just wish there was more data on how fast these games run on different CPUs because the 5900X seems to do a good chunk better than the 5600X in the Civ 6 Gathering Storm AI test but with only 2 reviewers that tested it (GN and OC3D) it is not a firm conclusion. Also the AMD vs Intel performance was mixed, OC3D had Intel/AMD neck and neck where as GN had AMD well ahead. Could be memory settings on that one but when GN did their memory tuning video they did not include Civ 6 AI turn time so you can't see what the gain will be from memory tuning in that game. Then you have the Paradox Grand Strategy games and nowhere tests them, or Cities Skylines and other large scale games that have a lot of AI running in the background so again there is no data to support a buying decision either way and like I said these games are the ones where spending the extra over the 5600x / 11600k for the 8c + parts might be worth it if you get tangible tic rate improvements.

The FPS are 'okay' on the laptop, details turned right down so it's perfectly playable, but obviously it's a much better experience playing on my main rig that's for sure. I use the YNAMP mod for the massive maps, which apparently should always crash late game, but to be fair they are usually okay for a good 500+ turns.

Like you these tests would be nice if they were done as Civ is still my most played game overall, so it's certainly important it runs well, but I know that it runs great on my 3900x and it's natural successor for me will be the 5900x most likely, so performance shouldn't regress anyway. I did run my memory at 3200CL14 for a while when I first installed the 3900x and bios support from ASRock for memory overclocking was a bit poor, but these days it's fine with 3600CL14, but without doing any proper tests I cant say if it makes any difference on Civ or anything else really. I can't say that I've noticed any improvement though, certainly nothing perceptible.

It's certainly a more difficult decision these days for Intel or AMD if you need to buy everything that's for sure, as the CPU choice/price for 5000 series is limited and poor... Intel seem the better value option ironically with AM4 socket due to retire, which obviously wont matter with Intel, as you'd need a new board regardless.