DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
If it's [H]ard, hopefully it still plugs in to the same socket.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
Broadwell-E vs Ryzen in games (PurePC)
- Thread starter -Sweeper_
- Start date
If it's [H]ard, hopefully it still plugs in to the same socket.
Well they're saying we have this socket till 2020
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,677
To call Skylake/Kabylake and specially SKL-EP/X for Sandy Bridge is like calling Zen for Bulldozer.
Ryzen now reached Sandy Bridge IPC while only being able to clock at ~4Ghz tops. It shows again how hard the IPC and clock game is. People are unrealistic if they think huge improvements are around the corner and its just a matter of "competition".
Actually, I just realized that you are bad at math too. The gap is 3% given that the performance is normalized to a 2500k as 100%.
Not that this really means anything. But if you're going to pick a nit, I'll do the same.
No. You originally said that it was "3.7% faster" and I replied saying that is not true. It doesn't matter which is the baseline that normalizes the scores, because one can compare directly the relative scores
1500X: 118.7
2600k: 115.7
118.7 / 115.7 = 1,025929127 ===> 2,592912705%
which I rounded to 2.6%, which can be further rounded to 3%.
2.6% and 3% are both correct, with one percentage being more precise than the other. However, your original "3.7%" is not correct.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
You missed the point. The values are *ALREADY* expressed as percentages normalized to the performance of the 2500k. Reread the graph. They aren't showing you FPS (otherwise, how would the 2500k get exactly 100?). They are showing you a percentage. That's the score.
115.7% to 118.7% = 3% difference.
I must have been smoking something to get 3.7 (I think I somehow thought the 115.7 was really 115.0 when I glanced at the graph). But you don't have to do any division there since the values are already percentages.
Last edited:
No. Math doesn't work that way. To know how much faster is one chip over the other you have to do
118.7% / 115.7% = 1,025929127
(the percents cancel in numerator and denominator) and 1,025929127 faster means 2.5929127% faster, which I rounded to 2.6%.
In this concrete case both values, the real 2.6% and your wrong 3%, are very close, and one can use one instead the other but in another case you would get the wrong conclusions. Consider game scores for i7 7700k and R7 1700. You would do
165.2% −128.4% = 36.8%
and believe that the KBL chip is about 40% faster in games. But in reality it is 'only' 29% faster
165.2 / 128.4 = 1.2866 ==> 28.66%
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
No. Because as Kyle has shown, Skylake/Kabylake is around a 20% IPC bump relative to SB. Whereas Bulldozer IPC was so atrocious that AMD is probably understating the IPC gain of 52%. Think about it. A 4.7/5GHz 9590 8 core CPU is utterly demolished by a 3.6/4.0GHz 8 core Zen, in many cases by well over 50%. Factor in the clock speed difference, and the IPC improvement is even better.
*in games. In applications, even hardware.fr shows it being much better. Of course, games matter more to the OC community. But still.
See, this is where I check out. Yes, Intel is better. I concede this readily. Don't buy a Ryzen if you want the best. But, that being said, you really have very little faith in engineers. Yes, they are having problems with 7nm. Yes, the end of silicon may soon be coming. And they've some pretty interesting challenges. But I'm a great believer in the motivating power of money. When money is on the line, I think Intel will deliver. I think Intel will deliver better than AMD. They have the resources and some of the best engineers in the world. One of their engineers is a personal friend of mine (I got to see the engineering sample of a 6 core mainstream part - I think it was a Coffeelake?). Anyway, I think they'll find a way out of this pickle we're in... so long as they think their ass is on the line.
Last edited:
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
The values are already percents. Therefore it is correct to say they are 3 percent(s) apart normalized to the 2500k. No division required.
Last edited:
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
Hardware.fr really is the best, though. Next to Kyle's work, of course. Too bad Google translate sucks ass, and I can't read French for sh*t.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
N/m I just realized you meant performance NOT normalized to the 2500k.
This is me today: https://en.wikipedia.org/wiki/Dunning–Kruger_effect
This is me today: https://en.wikipedia.org/wiki/Dunning–Kruger_effect
Last edited:
I natively read and write French and English equally well. They're OK and maybe lead their language market but .. I dunno, when it comes to hardware, it's always better to discuss it in English, French is just lacking.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
One up on me. I speak English, broken Spanish, and fluent Miami Spanglish (sh*t really is a language of its own at this point).
Comparing OC to OC, the 7700k is what, 30% faster than a 2600k? For 6 years, that's barely an improvement. I'm not here to argue the difficulty of increasing it, obviously it's hard, but let's not pretend Intel has made vast strides and that they've left Sandy Bridge in the dust. They've improved IPC like 20% and continuously increased the stock clocks while barely improving the OC'd speeds (and in some situations, reducing it from SB).
The outlook does change a bit when you start to look at the HEDT parts, but that wasn't what I was referencing. Intel adding mainstream 6 cores is when things will finally be a bit more enticing in that segment.
Slightly above Sandy in compute and slightly below in games. Overall it is Sandy-like.
Not true.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
Except not a good overclocker. So inferior to Sandy by a little bit.
Yes, you're right. I'm just very stupid today. This is a great example of when a person ought to STFU, because he's not very bright. Said person being me. Please accept my apologies.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
The last 6 years have been pathetic from a CPU standpoint.
Yeah. I'd have bought a 6 core Kaby Lake over Ryzen. I keep saying that, and some folks reply with "well it's not Intel's responsibility to make what you want, you just have to live with it." Except I'm the customer, damnit, and I don't think I'm the only one who'd buy this sh*t if they offered it, either. I mean, AMD may suck, and Ryzen may be a somewhat sh*ttier analogue of a 6 year old design. But it's still Intel's problem that Ryzen has any appeal at all, because they decided to leave a wide gap between mainstream and HEDT parts.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
As an aside, as Juanrga accidentally pointed out... the only reason to buy an AMD product on any kind of absolute performance level is if you are stupid and bad at math .
.To be honest, while I like the Ryzen CPU I wouldn't buy an AMD based system today. The motherboard platform is a mess because of AGESA / UEFI BIOS code but more than that, it's not as flexible as Intel's or as feature rich. It lacks the PCIe lane flexibility of Intel's Z270. The direct PCIe x4 link for M.2 doesn't provide a huge advantage there and any other M.2 drives would have to go over the PCIe lanes attached to the chipset, or get split on the dedicated interface reducing the advantage to nothing. Anything off the chipset has to go over a standard PCIe x4 link which is again, about like DMI 3.0. Again, no advantage. Then you have memory compatibility problems, less flexible options for SATA based RAID, M.2 RAID, etc. Basically from a storage or PCIe lane perspective AM4 gets its ass handed to it by Z270 and X99.
The platform isn't a total fail or anything, but choosing AM4 isn't without caveats. If you run simple storage configurations, Its sufficient but this is the [H] where we often have 8x SATA drives, more than 1 M.2 drive, dual GPU's etc. Like many other people I'm excited at the rumor of an HEDT type Ryzen setup on a better and more robust platform. That's worth looking at if AMD can get It right or to market quickly enough. The latter I have my doubts about.
The platform isn't a total fail or anything, but choosing AM4 isn't without caveats. If you run simple storage configurations, Its sufficient but this is the [H] where we often have 8x SATA drives, more than 1 M.2 drive, dual GPU's etc. Like many other people I'm excited at the rumor of an HEDT type Ryzen setup on a better and more robust platform. That's worth looking at if AMD can get It right or to market quickly enough. The latter I have my doubts about.
Pretty much. At the end of the day, single thread performance has increased so little, and the increasingly threaded nature of most applications means that an 8c/16t Sandy at 4GHz would, in fact, still be a good processor today. It seems we've hit a pretty hard wall when it comes to clock speeds and IPC, so it's hard to get super excited about any CPUs that have been released since Sandy.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
AMD will find a way to f*ck it up. You can take that to the bank.
I'd like to be able to argue against that point but that's hard to do. AMD's chipsets have always been mediocre at best. It's taken some serious work by motherboard manufacturers or chipsets from third parties like NVIDIA to bring the AMD platform into parity with Intel's at any point I can think of off hand.
jeremyshaw
[H]F Junkie
- Joined
- Aug 26, 2009
- Messages
- 12,511
Tellingly, Sony didn't use an AMD chipset for their PS4 southbridge. Analysis of the HW calls indicates Sony repurposed an existing ARM SoC for that.
Though, Sony is not really a shining example of what to do when it comes to implementation of HW.
Though, Sony is not really a shining example of what to do when it comes to implementation of HW.
AMD may not have ever had a chipset or the intention to build one for its console parts. They would have worked in conjunction with Sony to develop the CPU and interface with parts of the console already designed or being designed by other groups. It's not quite the same as designing CPU's for desktops in which the chipsets get produced to go with a CPU.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
And their drivers are almost as bad.
You are not wrong. I've had a fair amount of trouble with the SATA drivers on AM4.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
We can dream of the day when there is real competition for Intel again. Today is not that day. Tomorrow not looking so good either.
Meh.
I wouldn't agree with that necessarily. AMD isn't competition for the fastest system out there, but it has a place in the market. As I said, I think its a good alternative to Kaby Lake in a lot of areas. On the motherboard side, you may find the platform lacking in a more complex build. For a simple, streamlined setup its fine. Albeit the drivers and UEFI BIOS situation sucks but that will change in time.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
I'm pretty disappointed, overall. It's an 8 core chip, but it's gimped compared to the Broadwell-E platform. It's good for folks who want Broadwell-E throughput, but are broke (that'd be me, obviously). But you sacrifice gaming performance and feature set to get it, and you deal with AMD's typical driver and BIOS bullsh*t. It's a lot to give up. At first glance it looks great next to a $1050 Broadwell... but, as they say, there ain't no such thing as a free lunch.
Gideon
Supreme [H]ardness
- Joined
- Apr 13, 2006
- Messages
- 4,265
That is a poor assessment. Broadwell-E is a good chip no doubt and runs well, however it costs 1,000 bucks to have one. Very few people will spend that kind of cash on 1 part and [H] is not the best representation of the market. Ryzen will likely do very well in OEM machines and I think it will work great as a laptop cpu as well, it however is not the fastest cpu in all things. It does some things very well especially if it uses all the cores as AMD has made SMT work well. Were it does not do as well is single thread or just a couple threads and it shows. If your goal was to have the highest Benchmark machine then you chose poorly buying a AMD, it however is superb at some things and merely average at others. It's not the best gaming chip but I think you would be foolish to think it was going to be since very few games really make use of up to 16 threads. As for the motherboards they are fine for 98% of the market, less then 2% even use 2 or more gpu's anymore and NVME is still pretty new and has yet to really show much difference beyond synthetic against the SSD's. The boards have had Bios issues no doubt, but your buying anything when it first comes out on a whole new platform and process you better buckle up for a rough couple months for either chip maker. All in all everything works but sometimes it was frustrating getting there, the only fault you can give AMD on the bios side is the current memory support but they did not try to hide that fact and reviews showed the issue as well. If AMD built a chip that was faster then Broadwell you would know, it would cost the same or more, they priced where the average consumer will pick the 8 core over the Intel 4 core for the most part.
In the end it's a excellent chip for the money and considering how far they were behind it was a massive leap forward for them. People get far to tied up in benchmarks for their own good, if one get you 110 fps and the other is 100 fps it really does not matter except on a chart. Its fine to know which is faster but what is best for you at the price is what matters most.
Nonsense. Intel chipset based motherboards are usually pretty decent on launch. You don't have to strap in for a rough couple of months for most of them. AM4 is the biggest cluster fuck of a motherboard launch. In 20 years of working with hardware and over 10 years of reviews I've never seen a launch go this badly. I thought 990FX was a shit show but AM4 is on a new level.
Absolutely agreed, but I think some of that was board partners not taking this launch seriously either, though that's still on AMD for failing to give longer time lines to work with them.
DuronBurgerMan
[H]ard|Gawd
- Joined
- Mar 13, 2017
- Messages
- 1,341
And for AMD checking out of the market for so long. They lost a lot of cred by being out of the game. Maybe they'll get some back, now.
Half of this trouble is probably because AMD hadn't launched a decent CPU in so long, they basically forgot how to do it.
Gideon
Supreme [H]ardness
- Joined
- Apr 13, 2006
- Messages
- 4,265
Were just going to have to disagree on that, I am using it on the Crosshair VI and have had no issue other then my memory not being able to run at 3200, obviously you have forgotten about the Slot A days and the motherboards from VIA and AMD and others, that was a cluster fuck. I know you only buy the highest end crap and yeah a high end Intel motherboard has better features, but most people could care less about most of those features and they are expensive to add to the board. Ryzen feature set is the middle of the road, I am not shocked you think its so inferior based on how you buy stuff. But to say this is the worst launch ever, is just lunacy. Biggest issue has been keeping motherboard in stock for people to buy, which is finally starting to correct itself. Intel has had its share of issues as well.
https://www.extremetech.com/computi...es-intel-chips-to-freeze-in-complex-workloads
https://www.theregister.co.uk/2017/02/06/cisco_intel_decline_to_link_product_warning_to_faulty_chip/
http://blogs.intel.com/technology/2011/01/chipset_design_flaw/
A small sampling of Intel having issues, including one that had to be recalled with their chipset.
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,677
It didn't help that Ryzen was launched 3-6 months too early and its outsourced crappy AsMedia chipset.
With Intel its pretty much install and have fun. Not sitting with 117 issues of this and that.
Ryzen simply set a new record, and it wasn't a positive one. But that's what happens when the CEO demanded a certain launch date, ready or not.
With Intel its pretty much install and have fun. Not sitting with 117 issues of this and that.
Ryzen simply set a new record, and it wasn't a positive one. But that's what happens when the CEO demanded a certain launch date, ready or not.
I will sort of agree with you in part. No AM4 motherboard I've ever seen or read about thus far is on par with some of the worst motherboards I've seen in the past. If I did a top 5 worst motherboards of all time list, no AM4 motherboard would probably make the cut for that. At least, not so far. So I'll concede that I'm talking about general, platform / chipset specific issues. I still say that this is the worst, or at least is the worst one I've seen since I began reviewing hardware over 10 years ago.
What I said is nonsense is this:
I can't agree with this statement because it isn't true. While Intel has had some rough motherboard / product launches, it isn't a routine issue. You don't have to strap in for a rough couple of months most of the time. AMD product launches have been consistently worse. It's not always AMD's fault mind you, but on the platform side, AMD has almost always been worse off than Intel. Not only are they rarely comparable in features, but then you have driver issues and other, manufacturer specific problems due to the fact that AMD doesn't have the hold on the motherboard manufacturers that Intel does for ensuring quality products get built before they can throw their brand logos on the box.
Sure, you do take a risk being an early adopter and we all know that. So the risk of having to strap in for a couple of rough months is there but by in large you don't have to. Keep in mind that I've been using all of these chipsets across multiple brands, before launch now for a decade. So believe me, I see that shit before you guys do. I end up helping manufacturers work through that crap. I've seen many pre-release versions of motherboards that aren't ready for production.
This was only because AMD refused to delay Ryzen's launch for fear of undermining consumer confidence in the CPU itself, or at least that's my take on it. They'd rather fix the memory issues with AGESA code down the line than make it look like Ryzen had to be delayed because it was shit. That wouldn't have been reality, but for marketing people perception is reality. It isn't the informed people that shape public perception.
My buying the highest quality gear has nothing to do with anything. I work with, and use hardware across a vast range of price points and always have. I've been involved with computer hardware for two decades now. I've been a service center technician, in store computer technician, onsite computer technician and I've got many, many years in the IT industry. I've seen, built, and repaired thousands and thousands of machines. This has been going on since the mid-1990's. Believe me, I haven't forgotten any of the things you've mentioned. There are also two sides to a product launch, and not all of them are technical or visible to you guys. There are behind the scenes things you aren't privy to generally speaking. You guys learning about the AGESA code being released only three weeks before launch to manufacturers is the kind of thing you normally wouldn't learn about that I would. A lot of things don't get discussed with you, or on the forums. It's not necessarily that AM4 is the worst by itself, it's rather the combination of factors that led to the launch and the decisions surrounding them. It's also how AMD has handled things on the back end with manufacturers.
So you guys know, we've had multiple copies of some AM4 motherboards in house because we've had that many problems. I am not kidding when I said this is the worst launch I've ever seen. You may have had good luck with the Crosshair VI, but that doesn't mean it's problem free. You may simply have gotten lucky, or its your particular combination of hardware. You'll be reading more about this stuff as we start getting these things reviewed, but for us it's been rough dealing with AM4.
I really feel no one took this launch seriously and that's a reflection of AMD being nearly down and out of the builders CPU market. I bet the Asus FM4 guys felt like the guys designing boards for the VIA cpus sitting around thinking, "why did I get stuck with this crap, they should have stuck Zhang Wei with this job".
Everything going on regarding boards/microcode/RAM seems to be a reflection of only moderate interest and low manpower being assigned. I'm watching with keen interest to see how the prosumer line looks at launch.
Anyhow, looking forward to your findings regarding the boards.
Gideon
Supreme [H]ardness
- Joined
- Apr 13, 2006
- Messages
- 4,265
I still think your a little to easy on Intel and their issues, having to recall motherboards is a major fail to me and while this AM4 launch has been less then stellar, I just dont see it as too surprising as they just dont have the funds and staff to make it as smooth as most Intel launches. If Ryzen sells well and they can hire back staff then perhaps that will change and will get better launches, dont get me wrong I would enjoy a smoother launch as well. As for issues you guys have had, I am sure I have read about them on other forums and initial bioses were a mess for most motherboards and caused havoc. As for drivers I have not seen anything on that end being complained about so I am not sure what issue you have had there. I also dont disagree they should have held the launch back to fix these issues, but you push out a delay and some people on the fence will just go and buy something else instead. I also think they launched when they did cause more and more leaks were showing up and they had the performance to compete again and they wanted to capitalize on the bandwagon while they could. You should see some of the messes in the car industry with electronics and how borked things are at times, think people get pissed off when their computer is messed up, people are super pissed when their car is broken and you tell them you cant fix it until there is a update for the software released in a month or two. Sometimes you launch knowing it's less then ideal and just hope you can fix it as you go. As for my luck with AM4 most of it had to do with reading forums while I waited for my motherboard to arrive, plus I was a owner of the Slot A days so I am more then used to things being a pain in the ass. But I dont disagree with you that AMD should have had a better launch then what they ended up with.
eclap
n00b
- Joined
- Apr 15, 2017
- Messages
- 11
Yay, well done AMD, you kind of hover around a 5 year old Intel cpu performance. Congrats, pats on shoulders, job well done. Sorry, how is that an achievement?
Shintai
Supreme [H]ardness
- Joined
- Jul 1, 2016
- Messages
- 5,677
Its a big achievement when the buyer is vendor locked due to loyalty.