AMD Ryzen 9 5950X and 5900X faster than Intel at gaming?

So we finally have some third party reviews of the two most powerful mainstream CPUs from AMD. I wanted to share some benchs in gaming (in synthetic multicore benchmarks AMD wipes the floor with Intel):
























The 10900K has more room to overclock probably, and even stock vs stock it seems to be doing pretty well even tough it has 2 and 6 less cores than the 5950X and the 5900X respectively. What you guys think?

Source: https://www.extremetech.com/computing/316943-ryzen-9-5950x-and-5900x-review

 

I'd love to see some flight sim 2020 results. I'm glad I sold my 3900x now too, lol.

 

Found this one:



Is a german site and shows different settings for MS FS2020: https://tweakers.net/reviews/8270/1...ight-simulator-2020-en-crysis-remastered.html

It seems to be using a GeForce RTX 3090 also.

Are you going to get a 5900X?? or are you getting Intel now?

 

From the benchmarking results from the TechTubers 12hrs ago** it seems 5950X is the better of the bunch, however to be fair, 5900X results seems to be just a little shy when compared to 10900K. 5900X wins for being better value, but not "overwhelming performance" considering 5900X packs more core counts than the BlueChip.

**From Gamer Nexus, LTT, and HardwareUnboxed. Coincidentally, the 3 seems to upload their videos at the same time when i turn on my TV last night to see all 3 videos showing same uploading time stamp.

 

lol, probably for NDA they all three had to wait until a certain day time. I also watched the GN review. I think is a huge jump in gaming for AMD which used to be pretty much not a competitor for Intel at gaming. But being realistic AMD is just catching Intel in gaming, which of course is great. The 5950X should be a bit cheaper IMO though. The 10900K is being sold on eBay for $550-600. The 5950X should be no more than $700

 

The 10900K is easy to overclock and performs well. AMD really sucks at CPU and memory overclocking. If you're not into overclocking and generally just leave everything stock, the AMD CPU is probably the way to go if you don't mind spending more. If you enjoy overclocking, the AMD CPU is going to be about as exhilarating as watching paint dry.

 

I laughed way too hard at this....

 

Wrong - $579.99 versus $549.99 for the 5900X.

 

I stand corrected. Thank you. Everything else is accurate.

 

Intel's too hot for me in my higher than avg. ambient temps >35C.

 

Incorrect - $529.99 (which is the proper price) @ Best Buy:

https://www.bestbuy.com/site/intel-...ked-desktop-processor/6411492.p?skuId=6411492

But I'd still pony up the extra $20 for a 5900x since over the years I've regressed to a mild OC'er.

 

Not so fast. Depends if you can get one at MSRP

Scalpers Strike Again, Ryzen 5000 Processors Impossible to Find In stock tomshardware.com

Unfortunately, scalpers have once again stolen the show and bought as much volume as they can, and availability for any Ryzen 5000 series CPUs is almost impossible to find.

The scalpers are again, using eBay as their base of operations and have pushed prices of the 5000 series CPUs well over MSRP.

According to Ebay, you can buy the Ryzen 9 5950X for an absurd $1,699.99 from one of the several listings, "fortunately" most of the 5950X listings are more "reasonable" ranging between $1,050 and $1,200.

Stepping down to the Ryzen 9 5900X the most expensive listings touch the $1,000 mark while most hover in the $800-$900 range. The cheapest listing that we found comes in at $612 for a pre-ordered model. That's not too bad considering its only a $63 price hike.

 

In part I blame people who are buying at that prices. eBay is nothing else than supply and demand, if the price is doubling is because some people are buying at those prices. I been following RTX 3080 auctions on eBay and people are paying over $1K for the freaking card. $300 over the MSRP. I'm not paying a penny over over MSRP for the 3080 even if I have to wait to the middle of next year or more, lol

 

I mentioned it in another thread, but for many, even paying a thousand+ over MSRP is a nothing burger to them for many reasons and they have no problems paying those inflated prices. Like you said, supply and demand. I'll never pay over MSRP for anything and most times will wait to pay BELOW MSRP.

But if you have the funds, reasons or whatever and can't wait and want it NOW....well, go right ahead. I'll be waiting for you to regret it or moving onto the next big thing and buy it from you at a discount when all the hype dies down.

 

Yeah is up to debate, but, people are paying a lot over MSRP for no logical reason other than bragging rights. Personally I don't care about bragging rights I only care about functionality. In my case if I catch a 2080 Ti or a 2080 Super at a good price I'll get it and in the future I'll get a RTX 3080 or a 6800XT when prices come back to MSRP or lower. A well overclocked 2080 Ti can be as good as a stock RTX 3080. I mean is not like playing games with a 2080 Ti you'd be playing at bad frame rates, lol.

There is no reason to pay over MSRP for tech hardware from an economic point of view nor tech point of view, just from a psychological point of view it might worth.

 

It's OK guys. The way I looked at it, that simply wasn't worth debating so I just responded in a diplomatic manner.

I have seen both AMD and Intel CPUs selling for WAY MORE than the MSRP. That is super-stupid and it only happens because people are stupid and scalpers take advantage of stupid people. In the end, it hurts all of us whether we are stupid or just an unfortunate victim that ends up being adversely affected by the stupidity of those around us.

At times it is nearly impossible to buy one for the correct price. In the end it doesn't really change anything. I'm only going to buy what I want to buy and I will pay more for what I want rather than accept something I do not want simply because it is a better value in the opinion of some people. The way I see it, if a product doesn't do what I want it to do, then there is no logical reason to entertain the idea of buying it. Doing so would be a waste of money.

Removing the reference to price in the comments that I posted ultimately changes nothing. Intel overclocks CPU and RAM nicely and it is a fun product to own if you enjoy overclocking. AMD sucks at overclocking CPU and RAM (and GPU as well) and if you enjoy overclocking AMD is not the right brand for the job. It's that simple. If AMD products responded favorably to overclocking, there is a pretty decent chance I would want one. And, if I wanted it badly enough, I'd probably pay more than it is worth to get it. I wouldn't like that and it would tick me off, but that is better than the option of settling for something that is unable to meet my expectations based only on its price being less.

 

while the boys gets their hands on the multi cores mega chips...

...the subtile dutch lady gets her first contact with the subtile chip!​

 

What most these ppl doesn't realize or recognize, is how attention grabbing they're as well as how long does it take for a new designed/engineered/developed product to kick of the old one...and here I'm referring to one that's only just a year age gap.
The jump of desktop CPU model reference from 3xxx to 5xxx is just a... numbering thing from AMD. Certainly, and undeniably, there should be a performance increase to accompany with that leap of increment. Sellers scalp buyers because most of the time due to buyers stupidity of not realizing that manufacturing and production of the launched products is still going on in other regions where they are recovering from the pandemic situations... Why rush when there's going to be supplies from the factories tomorrow?

 

I’m not gonna argue that the 5900X and 5950X are both fast processors. But, some of the numbers from reviews are very wrong.

I saw a bone stock Ryzen 5950X beating a overclocked Intel 10980XE that was running 4.8Ghz on all cores. And thought, wow! That’s amazing! The 5950X is a monster.






Until I ran the test on my own 7980XE at 4.8Ghz which has slightly slower IPC than the 10980XE.


Here are my results. Not really sure what’s up with that review. But, it’s very misleading to a lot of people. And it makes me question the results in all of the testing.

 

Seems like TS doesn’t utilize all cores/threads.

 

So then why is my slower 7980XE 34% faster than their 10980XE in the same test, at the same frequency?

 

Seems you have missed the point. Different benchmarks can give different results. Remember also some benchmarks will benefits different with Quad vs Dual Channel ram.


Can you please run 3DM Firestrike and post your results?

 

The 10980XE has quad channel too though.

 

Forgot to add in that you have a unknown in the shared results... The different setups from different reviews use all sorts of ram, oc'd ram, different ram speed, timings etc A lot can affect the system performance.

 

This is absolutely right. RAM speeds, timings, ranks and channel mode can affect fps a lot. That makes a bit hard to compare an Intel CPU to an AMD CPU. Intel seems to react better to different RAM configuration than AMD

 

Apparently the new Ryzen 5000 series are showing a 10% performance uplift by using (2) sticks of DDR4 instead of (4) sticks.


PS, my DDR4 is decent. Maybe that’s it. But I was referring to the 10980XE at 4.8Ghz in the slide and not the 5950X

 

Odd, yet Intel chips show a slight performance boost (gaming at least) using 4 DIMMS vs 2 DIMMS.

 

Do you know if the settings are the same? That 16,388 score is much higher than everything on their chart..What does that CPU boost to normally?

Even if you had better RAM, there can't be a score difference of that much.

 

Here is +2000 points above. As I said in my posts, no point in speculate in the results or why the results don't match if you don't know the unknown. And ram can make an difference. Same with Os version, benchmark software version, drivers, bloat etc. Here 4.8 and ram at 3800MHz. Aka 2000 point higher than tps3443 but with "so called" slower ram. In exactly same way as I can beat many with slower sticks but with same/lower Cpu clock speed (4-6-8 cores processors). Too many variables to have the final answers.


https://www.3dmark.com/3dm/37271721

 

This is an interesting video. I follow this dude on YouTube and he is showing here the big gaming performance hit that a 10900K can get by using slow RAM sticks and not overclocking the ring ratio:

 

The jist I'm getting is that he disagrees with other YT tech reviewers trying to keep things as equal as possible with the out of the box experience and making sure memory is equal @ 3200 even as newer chips can take advantage of faster speed/settings.
You have to keep testing parameters equal for proper comparison to previous results and you can't go through and overclock everything as that introduces too many individualized situational variables. You also have to realize GN and similar are addressing the bulk of users and gamers not those who get in under the hood and tweak their configs.

No one is denying tweaking your system will net performance gains. That goes for both platforms.

He narrow focuses on Tomb Raider knowing games vary depending on builds and some games favor different configs differently than other. That's why YT tech reviewers run a wide swatch of games along with synthetics to get an overall average. I would have preferred to see him run a large suite of games and synthetics and then apply the same methodology (as much as possible) to a 5950x/5900x.

Never mind he's running his 10900k @ 5.4ghz and pushing everything. He does acknowledge his system and cooling and gear is an outlier at the end.

Then he starts extrapolating results into other gaming benchmarks.....that's where he loses me a bit.

In the end, what he wants to see is Zen3 tricked out vs 10th gen tricked out....

He will need to get his own 5950x/5900x to test the same memory at the same settings (or as high as it can be pushed on Zen3) while pushing Zen3 which he seems to be waiting on.

In the end, both chips can and will game like the beasts they are. Higher resolution, the less important till you hit 4k and in many games even a 9100f (lol) will net you the same performance.

His final Tomb Raider "findings" aka "Hey! If I push my settings, I get better performance"

 

Yeah the guy from that video (Jufes) was trying to show how gimped the 10900K can get by slow memories, his particular theory is that the 10900K scales better with memory speed than Zen 3 because the large cache that Zen 3 already has. 21% improvement just by rising the memory speed is a LOT. but you're right he is going to test several games once he gets his 5950X. Jufes has stated several times that he aims for the fastest gaming experience taking every component to it's limit, but he has said that out of the box the 5950X is a better option. GN is the one that does the better job when showing different charts with a lot of different chips and overclock settings, but Jufes has a point that at this day is unlikely to see a dude with a 5950X or a 10900K and 3200 MHz sticks, where for like $120 you can get 2x8GB 4400 MHz

 

Lol that was a terrible video. I just scrolled through it but I've learned not to trust this dude. Finding a result for Tomb Raider and applying it to other games is worthless. Intel's spec ram speed is 2933mhz so running tests at 3200mhz is already helpful. AMD ram speed matters (as does number of ram sticks just as it does in Intel). He sounds like an Intel fanboy that is sad because Intel lost the crown.

 

GN did a "super tuned" 5600x vs 10600 comparison:

 

Nice video there. The 5600X wiped the floor with the 10600K anyway tuning or no tuning . It seems that GN get a lot of comments regarding the 3200MHz memories they use in their regular benchmarks

 

Yeah, everybody has their moment to shine and right now AMD, overall, is top dog. Will Intel regain the crown next year? Who knows, but the most telling stat for me still is normalized @ 4ghz, clock for clock, IPC wise AMD is ~20% faster than Intel's chips atm.

I have no need to upgrade right now, but if I did, it would be AMD absolutely.

I'm most curious about the 5000+6000 numbers (along with Nvidia now working on their own SAM).

 

Great mashup of 1080p -> 4k results across the top two CPUs from both AMD and Intel:



Basically, 4k it really doesn't matter what you run from the top end CPUs. It is all equal or super close to not make much of a difference.

 

Yeah this CPU battle at gaming only makes sense at 1080p or lower. And out of the box AMD does a better job for their higher core count and better IPC performance. Things seems to get interesting for Intel only when overclocking and using fast RAMs to compete against AMDs large cache. Similar to what GN did.

By the way how you feel about Nvidia releasing a 3080 Ti for $999? same CUDA core count as the 3090, but 20 GB of GDDR6X @320 bit bus. Also Nvidia said they can also implement SAM in Ampere with any CPU, interesting times to come.

 

4K is so GPU bound even a 9100f is near equal depending on how multithreaded aware the game is at this point.

Nvidia working on their own implementation of SAM is great! It would neutralize one key selling point of RDNA2 if it pans out properly.

As for the 3080ti, that makes perfect sense. When Nvidia announced their lineup, we talked about it in the other threads. “Incoming 3080ti!” .



That cuda count is rumored at this point, but the real question for me is how many RTs? What about everything else still TBC? I suspect we’ll get more clarity after the 6800/6800xt launch this week what Nvidia plans to do. Early leaks show RDNA2 having potentially weak RT performance. I could see Nvidia releasing a 3080ti with equal or slightly less cuda cores than the 3090 with 20gb snd the same RT power as the 3080. The objective would be to match the 6900xt in rasterization and depending on leaks potentially keep everything else the same as a 3080.

Nvidia will do JUST enough to bring a 3080ti to market that will compete effectively enough with the 6900xt at the 1k price point. No more; no less. They don’t want to cannibalize their 3090 market but obviously will do what they need to do to equal or beat AMD at similar price points.

The beautiful thing is without AMD you maybe wouldn’t even get these options. Another thing I like that AMD did last cycle was offer RDNA, full power and basically say, “here you go. See you next product cycle for RDNA2.” I suspect they will do the same with RDNA2

Can’t wait for this week’s third party reviews of 6800/xt.

 

3200MHz indeed helps. I was astounded when I used 3200MHz Ryzen BGA U series and higher freq. did see a improvement in cpu/memory bound tasks. I reckon Intel will use 3200MHz as standard and afterwards they will be XMP2.0