I'm wary of non-Intel CPUs but I might try out Ryzen if I need a new CPU.
Going to get something of this caliber if the RTC 2080ti FTW3 I'm getting ends up being bottlenecked by my i5-6600 trying to get 144fps out of some games (mainly Apex which seems poorly optimized)
Intel has lately become increasingly deceptive about TDP, benchmarks, and has been cutting corners to improve IPC that resulted in the latest security vulnerabilities. Although AMD has arguably less successful branding, on an objective level why would you be wary of their CPUs any more than you would with Intel’s?
anandtech reported differences in their benchmark because of an updated BIOS version, is there any mention of what BIOS was used while benchmarking the 3900X?
Is it just me or is AMD seriously underpricing their processors? They could easily sell them for at least 20-30% more.
Yeah, it might be that Intel is overcharging for theirs, but they already set the expected prices, and AMD needs the cash, so why not sell for slightly higher prices?
I guess they want to maximize market share, consumer trust and brand awareness for the best long term gains. Hard to tell from my non existing experience, if someone with real world experience could chime in and give an educated guess.
AMD need to price competitively to gain market and thought share back e.g. (I'm guessing) Most data centres will be thinking of buying intel by default, then AMD just due to past experience.
The cost of switching is high for many companies. AMD has to price so competitively that they overcome that cost.
But also, Intel’s costs are going to be far higher since they’re using larger dies and aggressively binning. In contrast, with AMD’s smaller “chiplets” even the lower quality silicon (which they’re probably using in all the non-Epyc chips) can be utilized.
When I bought my first CPU for a PC build (an i7 930), I remember that the top of the line model (an i7 975) was in the $1,000 USD range.
According to an inflation calculator, there has been roughly a 20% inflation since 2009, so that price adjusted for inflation would be $1,200 USD for the top of the line processor.
However, the i9 9980XE is priced at $1,800 USD, a 50% increase.
I believe that rather than AMD underpricing, Intel has been marking up their prices since they have had a bigger market share and that $750 USD is a fair price for a top of the line AMD processor.
The 3900x is than half that adjusted price though and it's certainly not halfway in the product lineup. I mean yeah Intel has been steadily increasing the max price range for a while now but that doesn't change that this CPU is relatively cheap for where it is positioned (proud 3900X owner here).
The real question here (which we won't get an answer to) is who's margins are better... somebody else's 7nm + chiplet design certainly make it easier to manufacture.
I wonder whether it would be worth Intel lowering the price, while keeping the same production levels. Would the lost income cover the better marketing?
Also why didn't the article show the price difference given it is one of the most significant pluses of the Ryzen?
As exciting as this is, there’s some curious bits in here. Why is the 9980XE Cinebench score “simulated?” Why is AMD 64 and Intel 64 a separate feature? (Not to say there are zero differences but I don’t think there are many intentionally - maybe some ring 0 differences?)
It is also said there are “no reasons” to consider the 9980XE.. but I can see a few actually. It has more cores and threads, for starters.
I do agree the Ryzen here is a clear winner in value. It is Much cheaper even if it does not win in absolute performance.
Is it an advantage that the 9989XE has more cores and threads even if, according to the article, the Ryzen still wins in overall multithreaded performance?
No, but I am a bit perplexed by what that value entails. It looks like it is based on the Cinebench results, which, for the Intel chip, is “simulated.” There are probably also some workloads that behave differently than Cinebench.
On the other hand, I suspect the Ryzen 3950X should be a really good match when that arrives later this year.
There are many (small) differences between AMD64 and Intel 64. But it's obvious that the Intel processor implements the Intel 64 version of x64 and the AMD processor implements the AMD64 version. This also goes for VT-x, VT-d, and AMD-V.
Do most popular hypervisors support AMD-V in the same way that they support VT-x and VT-d?
For example with vmware player you can run a 64bit virtualized OS inside of an existing VM. Running a VM within a VM is quite useful in development sometimes.
The site quotes "40 watts lower power draw" as a reason to prefer the AMD CPU, but what they're actually comparing is TDP, Thermal Design Power. The TDP for Ryzen 3900X is noted as 125W, but it's actually 105W. According to Anandtech, the power threshold that is allowed to be delivered to the socket for 105W TDP CPUs is 142W.
The 3900X probably does use less power than the i9 9980XE, but TDP is not power usage.
But intel calculates their TDP at base clocks while AMD calculates theirs at boost clocks, so AMD definitely definitely has a better wattage per performance
According to tests by Tom's Hardware, the i9-9980XE draws 199w under full synthetic load and 245w in an AVX stress test at stock clocks. A comparison of TDP understates the efficiency advantage of the Ryzen chip.
Now I’m not following the cpu market anymore, but for as long as I remember (before y2k) AMD was always much more conservative than intel regarding the TDP.
I don’t know if it’s possible that in the last 10 years or so things changed drastically though.
They should specify what benchmarks and input sets were used and the benchmarking methodology. It'd be nice to also see the standard deviation plotted on these bars.
As a gamer? None of them really, the features table is also inaccurate on several rows. "AES" and "AES-NI" are the ame thing, and so are "Intel 64" and "AMD64". So it is a bit confusing to read because the features intersect more but the names are different on each row.
As a data analyst if you're using optimized libraries (numpy, Fortran or C/C++ compiled for your machine, manually written libraries), then you may care about AVX (advanced vector extensions), AVX2, and FMA (fused multiply add) operations.
I'm fine with autogenerated crap, (because frankly that's the only way I'll get a comparison between a g3258 and a ryzen 3600) but much less fine with incorrect autogenerated crap.
Interesting to note how AMD "coming back strong" has been received between their CPU and GPU lines.
On the GPU side Navi looked good, nVidia introduced their Super-RTX cards to compete, AMD dropped their prices before they were even released...
... To the casual observer, these are two peers slugging it out. With each news story or Use Case, maybe one is better for you.
In the CPU domain it's pretty universally "AMD wins" - "What will Intel do? We have no idea."
My entirely anecdotal feeling, is that the entire world has been waiting for a decade to give Intel a kicking - and finally with this release a huge stack of ammo has been delivered to pull the trigger on those pent-up-stories.
i.e. Bit that makes it interesting to me, is that whilst in the "GPU wars", Team Red and Green have had their fans on each side and seem to be duking it out - the commentary is generally "you the consumer are the winner"
In the CPU war, nobody is "Team Intel". Feels like the whole world just wants to pile on.
OK but which one has the most efficient backdoors? When the hidden TCP/IP stack sends all your data to the mothership, you don't want your connection to get too sluggish.
42 comments
[ 4.1 ms ] story [ 103 ms ] threadGoing to get something of this caliber if the RTC 2080ti FTW3 I'm getting ends up being bottlenecked by my i5-6600 trying to get 144fps out of some games (mainly Apex which seems poorly optimized)
There is nothing wrong with non-Intel CPU, but historically maybe some "considerations you might wish to make, when looking at the price"
Now.
No idea why you'd buy Intel.. None..
..You were right last year. Right now it's "it depends" (I'd call it a draw).
Yeah, it might be that Intel is overcharging for theirs, but they already set the expected prices, and AMD needs the cash, so why not sell for slightly higher prices?
But also, Intel’s costs are going to be far higher since they’re using larger dies and aggressively binning. In contrast, with AMD’s smaller “chiplets” even the lower quality silicon (which they’re probably using in all the non-Epyc chips) can be utilized.
According to an inflation calculator, there has been roughly a 20% inflation since 2009, so that price adjusted for inflation would be $1,200 USD for the top of the line processor.
However, the i9 9980XE is priced at $1,800 USD, a 50% increase.
I believe that rather than AMD underpricing, Intel has been marking up their prices since they have had a bigger market share and that $750 USD is a fair price for a top of the line AMD processor.
The real question here (which we won't get an answer to) is who's margins are better... somebody else's 7nm + chiplet design certainly make it easier to manufacture.
Also why didn't the article show the price difference given it is one of the most significant pluses of the Ryzen?
It is also said there are “no reasons” to consider the 9980XE.. but I can see a few actually. It has more cores and threads, for starters.
I do agree the Ryzen here is a clear winner in value. It is Much cheaper even if it does not win in absolute performance.
On the other hand, I suspect the Ryzen 3950X should be a really good match when that arrives later this year.
Seems niche but I guess it could happen?
For example with vmware player you can run a 64bit virtualized OS inside of an existing VM. Running a VM within a VM is quite useful in development sometimes.
The 3900X probably does use less power than the i9 9980XE, but TDP is not power usage.
https://www.anandtech.com/show/14605/the-and-ryzen-3700x-390...
https://www.tomshardware.com/reviews/intel-core-i9-9980xe-cp...
Difference of about ~57 watts (142 vs 199) in the non AVX.
As a data analyst if you're using optimized libraries (numpy, Fortran or C/C++ compiled for your machine, manually written libraries), then you may care about AVX (advanced vector extensions), AVX2, and FMA (fused multiply add) operations.
https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9980XE-v...
AMD is seriously dominating recently. Huge respect!
On the GPU side Navi looked good, nVidia introduced their Super-RTX cards to compete, AMD dropped their prices before they were even released... ... To the casual observer, these are two peers slugging it out. With each news story or Use Case, maybe one is better for you.
In the CPU domain it's pretty universally "AMD wins" - "What will Intel do? We have no idea."
My entirely anecdotal feeling, is that the entire world has been waiting for a decade to give Intel a kicking - and finally with this release a huge stack of ammo has been delivered to pull the trigger on those pent-up-stories.
i.e. Bit that makes it interesting to me, is that whilst in the "GPU wars", Team Red and Green have had their fans on each side and seem to be duking it out - the commentary is generally "you the consumer are the winner"
In the CPU war, nobody is "Team Intel". Feels like the whole world just wants to pile on.
WTF has the Intel PR dept been doing?
https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9980XE-v...
Suggest moving the discussion to Anandtech’s review or the review from another similarly high-quality publisher.