How to Save $6000 on a 28-core Flagship Intel Xeon: Platinum 8280 vs Gold 6258Rby Dr. Ian Cutress on August 7, 2020 8:00 AM EST
In February 2020, Intel launched eighteen new Xeon Scalable second generation processors. These mid-cycle additions to Intel’s product portfolio were designed to bolder up Intel’s server offerings on a very popular and very successful platform, adding in extra cores, extra frequency, or more cache than the previous offerings at roughly the same price. The goal of these ‘performance-oriented’ processors was to address customer requests in offering a more palatable performance-per-dollar offering. One of the new CPUs caught our eye: the Xeon Gold 6258R.
Skylake, Cascade Lake, Refresh
Colloquially known as ‘Cascade Lake Refresh’, these processors are the same silicon as the second generation Cascade Lake Xeon Scalable processors that were originally launched in April 2019. In most cases, the Refresh processors focus on both performance and performance-per-dollar metrics, especially given that Intel’s competition in this space were in a very competitive position and focusing on those values. Despite Intel’s data center revenue growing rapidly through 2019 and into 2020, there was a need to effectively replace or add new products into the areas where Intel believed it could keep a strong grasp on the customer base.
In our original announcement for the refresh parts, Intel touted an average performance gain of 36%, and a performance-per-dollar of 42%, although that was pictured as a 1st Gen to 2nd Gen Xeon Scalable jump. For a lot of the eighteen new processors on offer, they either added extra cores, more cache, or more frequency for the same cost as the parts they effectively replace. This usually comes with an increase in power consumption (there’s no escaping the physics), given there was no actual change to the underlying silicon, it simply was a function of binning and product margin.
One of the new parts was the Xeon Gold 6258R, with the R indicating ‘Refresh’. This processor was actually the highest core count refresh part, offering 28 cores at 2.7 GHz base and 4.0 GHz turbo within 205W.
For anyone who follows Intel’s server processor portfolio, those specifications look *very* familiar. Looking through the list, there is one very popular processor that has the exact same specifications: the Xeon Platinum 8280. Here’s the full breakdown:
|Intel 2nd Generation Xeon Scalable
|28 Cores / 56 Threads||Cores / Threads||28 Cores / 56 Threads|
|2700 MHz||Base Frequency||2700 MHz|
|4000 MHz||Turbo Frequency||4000 MHz|
|38.5 MB||L3 Cache||38.5 MB|
|3 x 10.4 GT/s||UPI Links||3 x 10.4 GT/s|
|8||Max Socket Suport||2|
|6 x DDR4-2933||DDR4 Support||6 x DDR4-2933|
|1 TB||DDR4 Capacity||1 TB|
|205 W||TDP||205 W|
The Platinum 8280 and the Gold 6258R are identical, almost to a fault. The same cores, the same frequency, the same power, and both support Optane DCPMM. The implementation difference is very subtle: where the 8280 supports 8-way socket deployments, the 6258R only supports 2-way. Intel has separated up the 8200 series and the 6200 series in this sole difference of socket support, which is actually more a firmware difference than anything else.
Oh, and the 6258R has a list price over $6000 cheaper.
Now, the reason why this is important comes down to where the 8280 sits in Intel’s Xeon portfolio. It is, for all intents and purposes, the processor that gets the most attention. It sits at the top of its public processor offerings*, it offers the most cores, and the list price is $10009**. If a non-technical executive is requesting ‘the best’ hardware for deployment, they naturally scroll to the most expensive part and add-to-basket. That processor would be the Xeon Platinum 8280.
However, most servers are single socket and dual socket, which essentially nullifies the ‘extra’ 4-socket and 8-socket capability that the Xeon 8280 offers. In this case, Does the 6258R, with the same specifications on core count, frequency, and power, perform the same as the 8280 but at a fraction of the price?
This is the question I set out to answer with access to both CPUs. Saving $6000 per single socket server, or $12000 in a dual socket configuration, would allow purchasers to focus that investment in other areas, such as memory or storage, or bring down the cost of purchasing quite considerably.
*Intel also offers a Xeon Platinum 8284 which also has the 28 cores that the 8280 does but is at a higher base frequency (same turbo) and 240W TDP. The list price is $15460, a +50% jump. This processor doesn’t seem to always be available everywhere, plus it was also launched months after the 8280.
**List prices from Intel are usually set as the price if someone buys 1000 units, so one would expect the individual cost would be slightly higher. However, major OEM partners and big hyperscalers rarely pay the list price, and the separate pricing is negotiated by contract. Rumors are that the big companies that might need a 100k units or more rarely pay more than 20-50%% of the list price. Exact figures are hard to come by.
Post Your CommentPlease log in or sign up to comment.
View All Comments
koguma - Friday, August 7, 2020 - linkYou're kind of comparing apples and oranges here. The threadripper is an enthusiast/workstation product. AWS isn't having any issues with virtualization using Epyc.
duploxxx - Friday, August 7, 2020 - linkNeither do we with our 1000+ amd servers when we switched from intel to epyc2 series. We have never seen some much price / perf / power ratio in favor of amd. Gone are all the security bugs and security measurements we had to take in our DC to not be confronted with the CVE. Our total server farm was halved and the total hw price was 2/3. With hypervisor it was also easy switch. Just select the maintenance window and switch.
Don't compare some workstation thread ripper virtualization who's with epyc2. There are a lot of DC moving into amd these days. Sure intel will still sell the bulk but the move by many was made on epyc2 and will increase with epyc3 launch. 2-3y ago when you ask oem consultant what hw to offer it was only intel. These days they ask what the server is used for and they offer depending on requirements.... Those who run intel only are dump old school that don't look at there budgets and performance and have no clue. Critical environment? Run on Z and you will find out what crytical really means.
ZoZo - Friday, August 7, 2020 - linkThat's great, but they've most certainly got the resources to tweak their software, with AMD engineers ready to assist. And they're probably not doing any PCI-E passthrough of USB controllers. Nor do the customers try to run virtualization software (Hyper-V, Windows Sandbox) within a Windows guest there.
Maybe it's just those "edge cases", but when you encounter problems with those, you start to wonder what else could go wrong in other cases.
duploxxx - Saturday, August 8, 2020 - linkYou are just working with a consumer playground. In server world we have dedicated clusters that run pci-e passthrough for nics and GPU. No issues on the vSphere at all. And Windows? What about it? +4000vm on windows would that be enough? With vm up to 60vcpu and 150tb hd space.
ZoZo - Saturday, August 8, 2020 - linkI specifically mentioned USB controllers because that FLR bug is on AMD's Starship and Matisse controllers regardless of platform. If those are the USB controllers on EPYC, then it will have the same problem (it locks up the PC, hard reboot required). Other things seem fine (GPU, NIC, NVMe SSD). The only "consumer playground" thing there is that perhaps you're more likely to be doing that on a consumer platform than on an enterprise server.
I didn't say Windows didn't run, I said nested virtualization didn't work. Please try to make an effort in reading comprehension.
duploxxx - Saturday, August 8, 2020 - linkThis is a thread about Server CPU. Not workstation. So why you bring this info than in the first place.... USB devices on server? You are joking right.
ZoZo - Sunday, August 9, 2020 - linkFine, focus on the USB thing and ignore the Windows nested virtualization. I guess some things don't (or didn't) just work on AMD workstation platforms, but everything has been absolutely peachy on server platforms.
Say company X needs new servers to run their complex IT infrastructure. But here's the thing: if something doesn't work, if there's an incompatibility, people you care about get executed. Oh, and you're going up against someone else doing the same thing, and if he gets it done faster, same punishment for you.
What would you go for? Intel or AMD?
Targon - Monday, August 10, 2020 - linkThat is why it takes several years for the server market to even consider making a shift or to start implementing a different design. Two years later, corporations have started to make the decision to test the waters with some Epyc servers and see how it goes. Considering the number of MAJOR security issues with Intel, the "tested" platform, any minor issues with Epyc won't be something to be concerned with.
TheJian - Saturday, August 8, 2020 - linkTHIS...The largest companies in the world are using "FAILED" AMD chips (sounds like you're saying they're failures yeeeeman!)...LOL. I haven't heard of amazon, facebook, google, microsoft etc, firing admins for buying EPYC. Intel pissed away 4.1B+ per year for 4.5yrs and now has fab issues because it should have been spent on 10/7nm (it was freaking 20B! That is a state of the art fab+ some). We are here now, because INTEL is exactly what you are claiming AMD is...trash right now. :) It happens to everyone at some point. Look at windows 10...ROFL. 10+ versions, and ALL SH!TE. They release a patch and it WTF's 4 versions of the OS...LOL.
How many bugs does an Intel chip have? SECURITY I mean?...Nuff said? We are talking their ENTIRE CHIP LIST for DECADES. Nah, Intel is 100% bug free right?...ROFLMAO. Caminogate. Timna, larrabee, TITANIC (LOL), etc etc. I could go on with the Intel failures but...whatever. They're the only other perfect thing every created on earth besides JESUS. I swear ;) They are the bees knees pal...LOL. That said, if AMD loses this time, it's because they refuse to CHARGE accordingly for their WINNING chips. If 4950x is really 500 (not $750 like the chip it replaces, rather pricing a 16c at 8c old pricing), again they will waste a top money maker. That is $250 off each chip sold. That is HUGE BANK to the NET INCOME line. AMD appears to be wasting a 4th gen on stupidity and pricing like a 2nd rate LOSER. Do you think Intel would have priced this at $4000 if you hadn't STUPIDLY priced your 64c at $7K? They're only 28c and they still are smart enough to charge MORE than 1/2 your 64c. Even in defense they make a right move if possible (get every dollar you can!). AMD would have priced a 28c vs. an intel 64c at $2000. That 28c was supposed to be around 17-25K and the 56c was supposed to be 50K. Then 64c AMD happened and NO NET INCOME for AMD, and still a RECORD for INTEL. You are NOT DOING IT RIGHT. Your stock after this Q report should have crashed $40 (and I mean $40 OFF the price, and even that is ridiculous, stock should be $20 with no improvement in INCOME). Quit giving away your silicon while Intel maximizes INCOME. The only point I really agree with the OP is you can't win like this. MAKE INCOME or SELL your company to someone who GETS HOW TO GROW. Lisa Sue is likely a great engineer, but management, well they need to show us the MONEY. Sue should be demoted to running divisions or something, not managing money for the company (or at least not firing whoever IS pricing products - surely she has some say here).
You are 20% of the market of x86 and make 600mil/yr. Intel owns the other 80% of x86 and still pulls down 23B NET INCOME (NOT REVENUE! that is like 70B+). Do the math, you should be 4B-5B NET INCOME at these levels or YOU ARE DOING IT WRONG (and that will still be a bit of a discount to win some share, but INCOME is more important). Your stock is ~1/2 of Intel at 100B (Intel is 208B Mkcap). YOU AMD, make 600mil/yr on that, they make 23B/YR TTM. Intel PE of 9, yours...165? LOL. You can live with a small share if you make NET INCOME. You can't live with a bunch of share but NO INCOME on it. Which, at 100-160mil/Q, you are making NOTHING NET. Sorry, 10% of what you should be making is NOTHING. Again, you idiots (you haven't learned in 4 gens if rumored prices are true for 4th ryzen), YOU ARE PRICING WRONG.
Your problem now? Intel will be buying YOUR WAFER STARTS, for MORE MONEY, so you can't now. Make hay while the sun shines...OOPS, someone forgot to learn this at AMD? Party is almost over, INtel contracts done or getting done for BILLIONS of TSMC wafers (180K, AMD only making 200K wafers from TSMC), and they still have fabs at 100% of their own. You are stupid. PERIOD. Four years WASTED on trying to get a few points of share rather than RAKING everyone you could for every dollar possible WHILE YOU CAN. Instead of the smart slower way, you took the quick price cut way and well, don't start wars you can't win yet. If you had priced everything within say 10% of Intel chips, they wouldn't even have cut ANY chip prices, as they have shareholders that require seeking NET INCOME and they keep doing it. AMD seems to refuse for some reason to attack correctly. Haven't read ART of WAR or Art of the Deal? It shows.
4am, no time to check if this is attack proof...LOL. To lazy to paste into word and this small box sucks. The data (math) is simple here. STock prices are not rocket science. SELL AMD ASAP, buy Intel, next xmas (2021) you will laugh at the ~26-27B NET INCOME they are making by then after adding 180K wafers to sell from TSMC. That is going to add to the bottom line. Intel is setting records and took a $27 hit to shares from $69. It won't take long for people to wake up and get that it is still making the same money as when it was $69 and heading higher NET INCOME wise. You'll be too late then ;) AMD is massively overpriced, so says NET, revenue, margin, mkap etc etc etc. When everyone loves your stock be afraid (yes I sold early :)), when everyone hates it, if metrics, fundamentals etc are all good (especially after FAKE NEWS crash over fabs that easily fixed by USING TSMC etc), BUY. Free money in 18mo. My family owns NO AMD (for now). I thought they'd be making BILLIONS at 20% share, but they forgot to price correctly for 4yrs. Oh well, dead money, or at least DANGEROUS for stockholders.
WraithR32 - Monday, August 10, 2020 - linkTL;DR