Ryzen 2 IPC improvements measured

[RotoCoreOne]

Just 2 more days for the embargo woooo~

[Mira Yurizaki]

4 minutes ago, 2Buck said:

How is it not "pretty nice"?

In the past, the average raw IPC performance bump between generations has been 10%-15%, save for two cases (Ceder Mill -> Core and Steamroller -> Zen) And Intel has more or less kept this trend and people think Intel's holding back.

 

So at best, I think it's average.

[Sakkura]

5 minutes ago, M.Yurizaki said:

In the past, the average raw IPC performance bump between generations has been 10%-15%, save for two cases (Ceder Mill -> Core and Steamroller -> Zen) And Intel has more or less kept this trend and people think Intel's holding back.

 

So at best, I think it's average.

This is essentially a tick, where Intel's IPC gains have ranged from literally zero to maybe 5-6% at best.

[JoeCoke]

10 minutes ago, M.Yurizaki said:

In the past, the average raw IPC performance bump between generations has been 10%-15%, save for two cases (Ceder Mill -> Core and Steamroller -> Zen) And Intel has more or less kept this trend and people think Intel's holding back.

 

So at best, I think it's average.

6700k -> 7700k had pretty much 0% IPC gain, and it's the same thing with 7700k -> 8700k. (This is just going off my memory + like 2 minutes of googling I just did, so I could be wrong, but in tests I saw when all these CPUs were at the same clock speeds they were the same.)

[Mira Yurizaki]

Just now, 2Buck said:

6700k -> 7700k had pretty much 0% IPC gain, and it's the same thing with 7700k -> 8700k. (This is just going off my memory + like 2 minutes of googling I just did, so I could be wrong, but in tests I saw when all these CPUs were at the same clock speeds they were the same.)

 

2 minutes ago, Sakkura said:

This is essentially a tick, where Intel's IPC gains have ranged from literally zero to maybe 5-6% at best.

Sure, if we look at Intel's last two recent offerings, but the complaint Intel has been sitting on their butts has been common since around Haswell. Or at the very least, had a lot of noise during leading up to Skylake.

 

But again, the historical average for IPC improvements has been 10%-15%. So Zen+'s improvements is more or less average.

[Nicnac]

As that POTUS would put it: Sad!

this would be ok for a generational improvement but they have to catch up or at least get much closer to intel first!

Now, don't get me wrong, I love AMD and would like to buy from them in the future. My first gaming rig was all AMD (phenom) and I am really hoping they can return to their former glory.

But I ain't getting my csgo frames this way! Pls do more AyMD

[JoeCoke]

12 minutes ago, M.Yurizaki said:

 

Sure, if we look at Intel's last two recent offerings, but the complaint Intel has been sitting on their butts has been common since around Haswell. Or at the very least, had a lot of noise during leading up to Skylake.

 

But again, the historical average for IPC improvements has been 10%-15%. So Zen+'s improvements is more or less average.

Of course it's average if we look at stuff from the past, 10% isn't that groundbreaking compared to past achievements. Pentium 4 to Core 2, lynnfield to sandy and Athlon XP to 64 are some that come to mind. But as far as recent times go, 10% (or ANY percent) improvement makes me happy. We've been stuck for quite a while.

 

It's just sad that it's IPC still hasn't quite caught up to Intel yet. Here's to hoping that Ryzen 2 will finally be 100% caught up or maybe, just maybe, no it's too crazy, be a little ahead.

[Suika]

Honestly, this is more than I anticipated. I assumed the majority of the benefits of Ryzen 2 would be higher clocks, I didn't know the improvements to cache would be so beneficial. I'm sure higher frequency memory could push it even further, if Ryzen 2 ends up having better compatibility.

 

I'm definitely considering the R7 2700X now, get an extra 500MHz, improved cache and latency, and possibly pick up a faster memory kit if the support is decent enough.

[The Benjamins]

6 minutes ago, Nicnac said:

As that POTUS would put it: Sad!

this would be ok for a generational improvement but they have to catch up or at least get much closer to intel first!

Now, don't get me wrong, I love AMD and would like to buy from them in the future. My first gaming rig was all AMD (phenom) and I am really hoping they can return to their former glory.

But I ain't getting my csgo frames this way! Pls do more AyMD

When was the last time Intel had a 15%-20% uplift in gaming?

[Coaxialgamer]

1 hour ago, Sakkura said:

I'm sure someone out there is annoyed that an improvement in cache and memory latency gets described as an IPC improvement. But whatever, the effect is the same.

 

If those numbers are legitimately the improvement at identical clocks, that's quite impressive. Along with a decent clock boost, Ryzen+ could be pretty awesome.

 

That old Ivy Bridge chip of mine is like...

 

  Reveal hidden contents

 

 

Kind of . In a sense the core isn't actually capable of higher throughput . 

But these improvements affect how much time your core will spend waiting around . You're decreasing the amount of NOP's and trading them in for actually useful work . It all depends if you're counting NOP's as "instructions" , but it's semantics anyway .

 

Plus , most IPC improvements today don't actually come from improvements to the core itself . They come from better branch prediction and better caching algorithms . The exynos M3 , nehalem , zen , ivy bridge and apple's hurricane all have wildly different "IPC" , but are all capable of the same theoretical throughput of 6 instructions per clock.

But again , it's semantics

 

 

[Mira Yurizaki]

Just now, The Benjamins said:

When was the last time Intel had a 15%-20% uplift in gaming?

Pretty recently, depending on the game.

[Nicnac]

Just now, The Benjamins said:

When was the last time Intel had a 15%-20% uplift in gaming?

you're totally right (unless we talking about ashes of the bench...). I was just trying to say that they aren't catching up as fast as I hoped they would. Still better than lazy-ass intel tho.

[Coaxialgamer]

1 minute ago, M.Yurizaki said:

Pretty recently, depending on the game.

Yes ,we did see those benefits , but they usually came about with higher clock speeds , not IPC improvements. 4790k to 6700k is a good example

[The Benjamins]

11 minutes ago, M.Yurizaki said:

Pretty recently, depending on the game.

wasn't that because they added 2 cores, and not because of clocks/ipc

 

10 minutes ago, Nicnac said:

you're totally right (unless we talking about ashes of the bench...). I was just trying to say that they aren't catching up as fast as I hoped they would. Still better than lazy-ass intel tho.

 

Keep in mind these tests is a underclocked 2700x vs a overclocked 1700

[Mira Yurizaki]

Just now, The Benjamins said:

wasn't that because they added 2 cores, and not because of clocks/ipc

Then be more specific next time.

[The Benjamins]

4 minutes ago, M.Yurizaki said:

Then be more specific next time.

I didn't think it was needed, this thread is about IPC improvements.

[mynameisjuan]

Reactions on LTT

 

Intel: "New processors have 10-15% IPC improvement- FUCK YOU INTEL AND YOUR MONEY GRABBING. GIVE US REAL IMPROVMENTS!

 

AMD: "New chips have 7% better IPC- HOLY SHIT THATS AMAZING! AMD IS THE BEST COMPANY EVER!!"

[Nicnac]

11 minutes ago, The Benjamins said:

 

 

 

Keep in mind these tests is a underclocked 2700x vs a overclocked 1700x

That does put it in a different light!

[SpaceGhostC2C]

32 minutes ago, M.Yurizaki said:

In the past, the average raw IPC performance bump between generations has been 10%-15%, save for two cases (Ceder Mill -> Core and Steamroller -> Zen) And Intel has more or less kept this trend and people think Intel's holding back.

Are you honestly telling me there was a 10-15% increase between generations since Sandy Bridge? In other words, are you telling me that Kaby Lake has between 60% and 100% higher IPC than Sandy Bridge (or 46-75% if we just don't count Broadwell because, well, Broadwell. In which case it would still be 60-100% for Coffee Lake vs. Sandy Bridge)?

 

I think you grossly overestimate it, but I'll be happy to see sources that support that claim.

I also think the 10% IPC claim for Ryzen+, while definitely "pretty good" if true, is not obvious looking at the available information. I could easily see Ryzen+ having a 10% or higher boost in performance, but IPC? Gaming benchmarks aren't the best tool for that.

[The Benjamins]

1 minute ago, Nicnac said:

That does put it in a different light!

ya, both where locked to 4ghz on all cores. even the 1800x didn't do that out the box, and the 2700x can boost 3.9 on all cores and 4.3 on 1.

 

these should compete against Intel a lot more in games.

[LAwLz]

Just wait for proper reviews... Not some random Youtuber that started making videos a month ago.

[ravenshrike]

 

33 minutes ago, M.Yurizaki said:

In the past, the average raw IPC performance bump between generations has been 10%-15%, save for two cases (Ceder Mill -> Core and Steamroller -> Zen) And Intel has more or less kept this trend and people think Intel's holding back.

https://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/9

No, just... no. There has been virtually nil IPC improvement since Skylake.There has been clock speed and memory speed improvement, but no(maybe 1% if you're being extra generous) IPC improvement since Skylake. The last IPC improvement to break 10% for Intel was between Ivy Bridge and Haswell.  Between Haswell and Skylake there was all of 5.7% IPC improvement.  The Ryzen 2 test posted was clock for clock with matched memory speed. Now, most of that 10% shown isn't IPC, it's slack being taken up everywhere else with memory timings on the CPU, probable adjustments to the infinity Fabric, and I/O. But 3% IPC gain AMD is actually claiming on a half(at best) node shrink is certainly nothing to scoff at.

 

[Sakkura]

52 minutes ago, M.Yurizaki said:

 

Sure, if we look at Intel's last two recent offerings, but the complaint Intel has been sitting on their butts has been common since around Haswell. Or at the very least, had a lot of noise during leading up to Skylake.

 

But again, the historical average for IPC improvements has been 10%-15%. So Zen+'s improvements is more or less average.

No, it goes just as much for ticks earlier on. Like Sandy -> Ivy Bridge. The IPC gain there was barely 5% (and that was enough for Intel to call it a tick+ rather than a "normal" tick).

[Drak3]

38 minutes ago, mynameisjuan said:

Reactions on LTT

 

Intel: "New processors have 10-15% IPC improvement- FUCK YOU INTEL AND YOUR MONEY GRABBING. GIVE US REAL IMPROVMENTS!

 

AMD: "New chips have 7% better IPC- HOLY SHIT THATS AMAZING! AMD IS THE BEST COMPANY EVER!!"

Intel hasn't improved IPC though, just clockspeed and higher base memory speed. Clock for clock, the 6700K, 7700K, and 8700K all have the same single core and quad core performance.

Whereas the 2700X seems to have an edge over the 1700X clock for clock.

[Camoxide]

I'm surprised there was any gains to be honest considering this is still based on 'Zen 1', the really improvements should be next year with Ryzen 3/Zen 2